Problems with Dosemu-1.1.3-Patchset 5

Problems with Dosemu-1.1.3-Patchset 5

[J.Strohschnitter]

Hi there,

I've got a crash of my dosemu while starting a printjob from our application.
Some other printjobs prints wiithout any problem.

I've started the dosemu with the -o option to log the error:

ERROR: cpu exception in dosemu code outside of VM86()!
trapno: 0x0e  errorcode: 0x00000006  cr2: 0x73fd0204
eip: 0x401ace06  esp: 0xbffff8c4  eflags: 0x00010602
cs: 0x0023  ds: 0x002b  es: 0x002b  ss: 0x002b
Page fault: write instruction to linear address: 0x73fd0204
CPU was in user mode
Exception was caused by non-available page
  VFLAGS(b): 010000011000000010
EAX: 73f9b9d4  EBX: 00034830  ECX: 00000001  EDX: 6bc6ea0d  VFLAGS(h): 
00010602
ESI: 083617f7  EDI: 73fd0204  EBP: bffff8cc
CS: 0023  DS: 002b  ES: 002b  FS: 0000  GS: 0000
FLAGS: IF DF RF  IOPL: 0
OOPS : 89 f9 83 e1 03 29 c8 4f 4e fd -> f3 a4 fc 83 c7 fd 83 c6 fd 89
leavedos(4|0x4) called - shutting down

Can anyone help me ?


My environment:

Dosemu 1.1.3 Patch 5
Linux SuSE 7.2 Kernel 2.4.16
Dosemu includes Novell-Networking


-- 

SET THE CONTROLS 
	FOR THE HEART OF THE SUN

Jens

Re: Problems with Dosemu-1.1.3-Patchset 5

[Bart Oldeman]

On Fri, 11 Oct 2002, J.Strohschnitter wrote:

> I've got a crash of my dosemu while starting a printjob from our application.
> Some other printjobs prints wiithout any problem.
>
> I've started the dosemu with the -o option to log the error:
>
> ERROR: cpu exception in dosemu code outside of VM86()!
> trapno: 0x0e  errorcode: 0x00000006  cr2: 0x73fd0204
> eip: 0x401ace06  esp: 0xbffff8c4  eflags: 0x00010602

could you check dosemu.map to see where 0x401ace06 is?
It might just be out of DOSEMU's scope; in that case it might be a DPMI
page fault and I can't really help you (Maybe Stas can).

dosemu.map is in the bin directory upon compilation.

Bart

Re: Problems with Dosemu-1.1.3-Patchset 5

[Stas Sergeev]

Hello.

Bart Oldeman wrote:
> could you check dosemu.map to see where 0x401ace06 is?
> It might just be out of DOSEMU's scope; in that case it might be a DPMI
> page fault
Current buggy fault handler (to be fixed
with my fault handling patch) have this:
if (in_dpmi)
   return dpmi_fault(scp).

So, while it is a bug, if DPMI is active,
there seems to be no way to reach that msg
at all.
So I would say this is a buffer overflow most
likely, and DPMI is disabled. gdb might help.
This all will be much easier if I get around
to implement the automatic gdb dispatcher,
but for now you told there is another buffer
overflow in mfs, isn't it?

Re: Problems with Dosemu-1.1.3-Patchset 5

[Bart Oldeman]

On Tue, 15 Oct 2002, Stas Sergeev wrote:

> Bart Oldeman wrote:
> > could you check dosemu.map to see where 0x401ace06 is?
> > It might just be out of DOSEMU's scope; in that case it might be a DPMI
> > page fault
> Current buggy fault handler (to be fixed
> with my fault handling patch) have this:
> if (in_dpmi)
>    return dpmi_fault(scp).
>
> So, while it is a bug, if DPMI is active,
> there seems to be no way to reach that msg
> at all.
> So I would say this is a buffer overflow most
> likely, and DPMI is disabled. gdb might help.
> This all will be much easier if I get around
> to implement the automatic gdb dispatcher,
> but for now you told there is another buffer
> overflow in mfs, isn't it?

No, that's what *you* suspected. I simply don't know. But if it would
crash in the MFS code the page fault would occur for a different eip (at
least with my dosemu.map).

We'll see.

Bart

Re: Problems with Dosemu-1.1.3-Patchset 5

[Stas Sergeev]

Hello.

Bart Oldeman wrote:
>> but for now you told there is another buffer
>> overflow in mfs, isn't it?
>  No, that's what *you* suspected.
I was referring to this:
---
Silly buffer overflow
indeed but I never imagined network redirectors working in 
such a way
---
Yes, this is fixed in "Patchset 5", but I
just doubted that .5 is already available
for somebody else. Seems like it is, though.

> We'll see.
Why do you prefer .map lookup rather than a
gdb stack trace? When DPMI is not used, gdb
is quite helpfull (unless instremu is used,
but it isn't).

Re: Problems with Dosemu-1.1.3-Patchset 5

[Bart Oldeman]

On Tue, 15 Oct 2002, Stas Sergeev wrote:

> Bart Oldeman wrote:
> >> but for now you told there is another buffer
> >> overflow in mfs, isn't it?
> >  No, that's what *you* suspected.
> I was referring to this:
> ---
> Silly buffer overflow
> indeed but I never imagined network redirectors working in
> such a way

I understand ... the poster with this bug is the same person whom I sent
the bug fix to the buffer overflow to. So I'm assuming that he applied
that patch.

> > We'll see.
> Why do you prefer .map lookup rather than a
> gdb stack trace? When DPMI is not used, gdb
> is quite helpfull (unless instremu is used,
> but it isn't).

One small step at a time.

Bart

Re: Problems with Dosemu-1.1.3-Patchset 5

[J.Strohschnitter]

> On Fri, 11 Oct 2002, J.Strohschnitter wrote:
> > I've got a crash of my dosemu while starting a printjob from our
> > application. Some other printjobs prints wiithout any problem.
> >
> > I've started the dosemu with the -o option to log the error:
> >
> > ERROR: cpu exception in dosemu code outside of VM86()!
> > trapno: 0x0e  errorcode: 0x00000006  cr2: 0x73fd0204
> > eip: 0x401ace06  esp: 0xbffff8c4  eflags: 0x00010602
>
> could you check dosemu.map to see where 0x401ace06 is?
> It might just be out of DOSEMU's scope; in that case it might be a DPMI
> page fault and I can't really help you (Maybe Stas can).
>
> dosemu.map is in the bin directory upon compilation.
>
> Bart

Hi,

I've tried to solve my problem, by contacting the Dosemu-Mailinglist and
Bart Oldeman - but both have no solution. So I hope you can help me by 
solving this problem.

I've checked out the dosemu.map but there is no entry with the error-codes
listed on the error-message.

Do you know what kind of error this could be ? Is it an dosemu-problem ?
The first message that will  be displayed (and is not logged in the file) is,
that the network-connection timed out and all mapped drives are not availabel.
So the application crashes and after that, dosemu crashes too and loggs the
message [RROR: cpu exception in dosemu code outside of VM86()! ....].



-- 

SET THE CONTROLS 
	FOR THE HEART OF THE SUN

Jens

Re: Problems with Dosemu-1.1.3-Patchset 5

[J.Strohschnitter]

> On Fri, 11 Oct 2002, J.Strohschnitter wrote:
> > I've got a crash of my dosemu while starting a printjob from our
> > application. Some other printjobs prints wiithout any problem.
> >
> > I've started the dosemu with the -o option to log the error:
> >
> > ERROR: cpu exception in dosemu code outside of VM86()!
> > trapno: 0x0e  errorcode: 0x00000006  cr2: 0x73fd0204
> > eip: 0x401ace06  esp: 0xbffff8c4  eflags: 0x00010602
>
> could you check dosemu.map to see where 0x401ace06 is?
> It might just be out of DOSEMU's scope; in that case it might be a DPMI
> page fault and I can't really help you (Maybe Stas can).
>
> dosemu.map is in the bin directory upon compilation.
>
> Bart

Hi,

I've checked out the dosemu.map but there is no entry with the error-codes
listed on the error-message.

Do you know what kind of error this could be ? Is it an dosemu-problem ?
The first message that will  be displayed (and is not logged in the file) is,
that the network-connection timed out and all mapped drives are not availabel.
So the application crashes and after that, dosemu crashes too and loggs the
message [RROR: cpu exception in dosemu code outside of VM86()! ....].


-- 

SET THE CONTROLS 
	FOR THE HEART OF THE SUN

Jens