[Qemu-devel] Re: Bug in emulation of 'bound' x86 instruction?

[Qemu-devel] Re: Bug in emulation of 'bound' x86 instruction?

[Frode Vatvedt Fjeld]

Frode Vatvedt Fjeld <frodef@cs.uit.no> writes:

> I'm suspecting that there's a bug in Qemu's emulation of the x86
> 'bound' instruction. The effect of this bug seems to be to add 1 to
> the ESP register, which of course havocs everything. [..]

Well, the response to this report has been nothing, but I'd really
like to see it fixed so that I'll be able to use qemu, which otherwise
seems to be really great. Also, it seems to me that if my report is
correct then this bug is quite serious, and can affect a lot of
software in unpredictable ways. I don't think I'll be able to submit a
patch for it myself, but if I can somehow assist someone "in the know"
with anything, please let me know.

Regards,
-- 
Frode Vatvedt Fjeld

[Qemu-devel] Re: Bug in emulation of 'bound' x86 instruction?

[Stefan Kisdaroczi]

Hi all,

While testing my "DOS+Realtime Multitasking Extension" Images ( working 
with bochs ), two things did not work with qemu :

1) The BOUND instruction seems to raise always the Exception 5
if the second Parameter is a location in the cs seg.
Example: bound di, cs:1095

2) An Exception 13 is raised when a "les" instruction should get the 
seg:ofs pair from the ss seg.
Example: les di, ss:si+10

-> im not really sure if its really dependent of the used seg, but i 
have started multiple programs, and all crash on different locations 
always on the bound or les instructions. It was always cs used in case 
1) and ss used in case 2).

The system is running in mixed 16/32 Bit Protected mode. It happens in a 
16-bit code segment.

The System debugger starts automatically on the exceptions 5 and 13, so 
i have looked at the memory locations referred and the values there are 
correct.

What can i do to help you helping me ?

Thank you.

Greetings kisda

[Qemu-devel] Re: Bug in emulation of 'bound' x86 instruction?

[Frode Vatvedt Fjeld]

Fabrice Bellard <fabrice@bellard.org> writes:

> I just fixed the bug you mentionned regarding the 'bound'
> instruction. Any code using the bound instruction was likely to
> fail, so this fix may allow new OSes or programs to run...

That's great, thank you.

-- 
Frode Vatvedt Fjeld