Re: [ros-svn] [blight] 18292: Huh, what? Oops! Just some stuff which noone cares about...

[Hartmut Birr]

[e-mail] wrote:
> Huh, what? Oops! Just some stuff which noone cares about...
>
> Updated files:
> trunk/reactos/ntoskrnl/include/internal/i386/fpu.h
> trunk/reactos/ntoskrnl/ke/i386/exp.c
> trunk/reactos/ntoskrnl/ke/i386/fpu.c
>
> Ros-svn mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-svn

Something is wrong in this changes. If I try to install HDD-Health, I 
get a crash.

- Hartmut

Assertion (Ke386GetCr0() & X86_CR0_TS) == 0 failed at 
ntoskrnl\ke\i386\fpu.c:337
A problem has been detected and ReactOS has been shut down to prevent 
damage to your computer.

Technical information:

*** STOP: 0x00000000 (0x00000000,0x00000000,0x00000000,0x00000000)

Frames:
<ntoskrnl.exe:1c76 (ntoskrnl/ke/bug.c:487 (KeBugCheckEx@20))>
<ntoskrnl.exe:1c8d (ntoskrnl/ke/bug.c:503 (KeBugCheck@4))>
<ntoskrnl.exe:556b6 (ntoskrnl/ke/i386/fpu.c:337 (KiGetFpuState))>
<ntoskrnl.exe:545c1 (ntoskrnl/ke/i386/exp.c:904 (KeTrapFrameToContext@12))>
<ntoskrnl.exe:54bab (ntoskrnl/ke/i386/exp.c:1195 (KiDispatchException@20))>
<ntoskrnl.exe:577ff (ntoskrnl/ke/i386/usertrap.c:133 
(KiUserTrapHandler@12))>
<ntoskrnl.exe:5bcc9 (ntoskrnl/mm/i386/pfault.c:132 (KiPageFaultHandler))>
<ntoskrnl.exe:57251 ({standard input}:152 (KiTrapProlog2))>
<shell32.dll:ad18 (lib/shell32/shell32_main.c:480 (SHGetFileInfoW@20))>

[Richard Campbell]

Evidently SOMEONE cared about it ;)

Hartmut Birr wrote:
> [e-mail] wrote:
> > Huh, what? Oops! Just some stuff which noone cares about...
> >
> > Updated files:
> > trunk/reactos/ntoskrnl/include/internal/i386/fpu.h
> > trunk/reactos/ntoskrnl/ke/i386/exp.c
> > trunk/reactos/ntoskrnl/ke/i386/fpu.c
> >
> > Ros-svn mailing list
> > [e-mail]
> > http://www.reactos.org/mailman/listinfo/ros-svn
>
> Something is wrong in this changes. If I try to install HDD-Health, I 
> get a crash.
> Hartmut
>
> Assertion (Ke386GetCr0() & X86_CR0_TS) == 0 failed at 
> ntoskrnl\ke\i386\fpu.c:337
> A problem has been detected and ReactOS has been shut down to prevent 
> damage to your computer.
>
> Technical information:
>
> *** STOP: 0x00000000 (0x00000000,0x00000000,0x00000000,0x00000000)
>
> Frames:
> <ntoskrnl.exe:1c76 (ntoskrnl/ke/bug.c:487 (KeBugCheckEx@20))>
> <ntoskrnl.exe:1c8d (ntoskrnl/ke/bug.c:503 (KeBugCheck@4))>
> <ntoskrnl.exe:556b6 (ntoskrnl/ke/i386/fpu.c:337 (KiGetFpuState))>
> <ntoskrnl.exe:545c1 (ntoskrnl/ke/i386/exp.c:904 
> (KeTrapFrameToContext@12))>
> <ntoskrnl.exe:54bab (ntoskrnl/ke/i386/exp.c:1195 
> (KiDispatchException@20))>
> <ntoskrnl.exe:577ff (ntoskrnl/ke/i386/usertrap.c:133 
> (KiUserTrapHandler@12))>
> <ntoskrnl.exe:5bcc9 (ntoskrnl/mm/i386/pfault.c:132 (KiPageFaultHandler))>
> <ntoskrnl.exe:57251 ({standard input}:152 (KiTrapProlog2))>
> <shell32.dll:ad18 (lib/shell32/shell32_main.c:480 (SHGetFileInfoW@20))>
>
> Ros-dev mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-dev

[Gregor Anich]

> Evidently SOMEONE cared about it ;)

Because it broke something, which wasn't my intention.

[Casper Hornstrup]

> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]] On Behalf Of Gregor Anich
> Sent: 6. oktober 2005 20:52
> To: [e-mail]
> Subject: Re: [ros-dev] Re: [ros-svn] [blight] 18292: Huh,what? Oops! Just some stuff which noone
> cares about...
> > Evidently SOMEONE cared about it ;)
>
> Because it broke something, which wasn't my intention.

It never is ;-)

[Richard Campbell]

Not to nitpick, but you should always describe what you are doing in the 
commit messages.  That's what they are there for...sometimes your code 
will be wrong and developers have tried what you have tried before.  
They then see what you are doing and explain why it's wrong.  Not 
everyone looks at patches.

Casper Hornstrup wrote:
> > -----Original Message-----
> > From: [e-mail] [mailto:[e-mail]] On Behalf Of Gregor Anich
> > Sent: 6. oktober 2005 20:52
> > To: [e-mail]
> > Subject: Re: [ros-dev] Re: [ros-svn] [blight] 18292: Huh,what? Oops! Just some stuff which noone
> > cares about...
> > > Evidently SOMEONE cared about it ;)
> >
> > Because it broke something, which wasn't my intention.
>
> It never is ;-)
>
> Ros-dev mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-dev

[Gregor Anich]

On Thursday 06 October 2005 17:49, Hartmut Birr wrote:
> [e-mail] wrote:
> > Huh, what? Oops! Just some stuff which noone cares about...
> >
> > Updated files:
> > trunk/reactos/ntoskrnl/include/internal/i386/fpu.h
> > trunk/reactos/ntoskrnl/ke/i386/exp.c
> > trunk/reactos/ntoskrnl/ke/i386/fpu.c
> >
> > Ros-svn mailing list
> > [e-mail]
> > http://www.reactos.org/mailman/listinfo/ros-svn
>
> Something is wrong in this changes. If I try to install HDD-Health, I
> get a crash.
> Hartmut
>
> Assertion (Ke386GetCr0() & X86_CR0_TS) == 0 failed at
> ntoskrnl\ke\i386\fpu.c:337
> A problem has been detected and ReactOS has been shut down to prevent
> damage to your computer.
>
> Technical information:
>
> *** STOP: 0x00000000 (0x00000000,0x00000000,0x00000000,0x00000000)
>
> Frames:
> <ntoskrnl.exe:1c76 (ntoskrnl/ke/bug.c:487 (KeBugCheckEx@20))>
> <ntoskrnl.exe:1c8d (ntoskrnl/ke/bug.c:503 (KeBugCheck@4))>
> <ntoskrnl.exe:556b6 (ntoskrnl/ke/i386/fpu.c:337 (KiGetFpuState))>
> <ntoskrnl.exe:545c1 (ntoskrnl/ke/i386/exp.c:904 (KeTrapFrameToContext@12))>
> <ntoskrnl.exe:54bab (ntoskrnl/ke/i386/exp.c:1195 (KiDispatchException@20))>
> <ntoskrnl.exe:577ff (ntoskrnl/ke/i386/usertrap.c:133
> (KiUserTrapHandler@12))>
> <ntoskrnl.exe:5bcc9 (ntoskrnl/mm/i386/pfault.c:132 (KiPageFaultHandler))>
> <ntoskrnl.exe:57251 ({standard input}:152 (KiTrapProlog2))>
> <shell32.dll:ad18 (lib/shell32/shell32_main.c:480 (SHGetFileInfoW@20))>

Oh, uh, oops!
I guess thats from a SMP machine? If it was SMP, can you try with a single CPU 
and let me know if it still crashes? (Sorry, I didnt think of SMP)

To temporary fix the problem you can make KiGetFpuState return NULL always I 
think.
Unfortunately I dont have a SMP machine to test so maybe it will take a bit 
until I find a good solution.

[Hartmut Birr]

Gregor Anich wrote:
> Oh, uh, oops!
> I guess thats from a SMP machine? If it was SMP, can you try with a single CPU 
> and let me know if it still crashes? (Sorry, I didnt think of SMP)
>
> To temporary fix the problem you can make KiGetFpuState return NULL always I 
> think.
> Unfortunately I dont have a SMP machine to test so maybe it will take a bit 
> until I find a good solution.

It is from a UP machine.

- Hartmut

[Gregor Anich]

> It is from a UP machine.

Please try this patch.

[Hartmut Birr]

Gregor Anich wrote:
> > It is from a UP machine.
>
> Please try this patch.

The patch does fix my problem. I didn't see this problem on the smp 
machine. I think the reason is, that the fpu state is always saved on a 
context switching.

- Hartmut

[Gregor Anich]

On Thursday 06 October 2005 22:13, Hartmut Birr wrote:
> Gregor Anich wrote:
> > > It is from a UP machine.
> >
> > Please try this patch.
>
> The patch does fix my problem. I didn't see this problem on the smp
> machine. I think the reason is, that the fpu state is always saved on a
> context switching.
> Hartmut

Yes, you are likely right. Maybe I will change the SMP stuff one day and make 
it a bit more efficient.

Thanks for testing the patch, and of course also for finding this bug!