XPost: comp.theory
From: NoOne@nospicedham.NoWhere.com
On 12/22/2020 11:40 AM, Robert wrote:
> In comp.lang.asm.x86 olcott wrote in part:
>> On 12/22/2020 4:07 AM, Bernhard Schornak wrote:
>>> olcott wrote:
>>>
>>> _H_Hat:
>>> [000005e6](01)? 55????????????????? push ebp
>>> [000005e7](02)? 8bec??????????????? mov ebp,esp
>>> [000005e9](01)? 51????????????????? push ecx
>>> [000005ea](03)? 8b4508????????????? mov eax,[ebp+08]
>>> [000005ed](01)? 50????????????????? push eax
>>> [000005ee](03)? 8b4d08????????????? mov ecx,[ebp+08]
>>> [000005f1](01)? 51????????????????? push ecx
>>> [000005f2](05)? e8effdffff????????? call 000003e6
>>> [000005f7](03)? 83c408????????????? add esp,+08
>>> [000005fa](03)? 8945fc????????????? mov [ebp-04],eax
>>> [000005fd](04)? 837dfc00??????????? cmp dword [ebp-04],+00
>>> [00000601](02)? 7404??????????????? jz 00000607
>>> [00000603](02)? ebfe??????????????? jmp 00000603
>>> [00000605](02)? eb01??????????????? jmp 00000608
>>> [00000607](01)? f4????????????????? hlt
>>> [00000608](02)? 8be5??????????????? mov esp,ebp
>>> [0000060a](01)? 5d????????????????? pop ebp
>>> [0000060b](01)? c3????????????????? ret
>>>
>>>
>>> 0603 jumps to itself. Reduce to
>>>
>>> _H-Hat:movl? 0x0C(%esp), %eax
>>> ?????? subl? $0x08,????? %esp
>>> ?????? movl? %eax,?????? 0x00(%esp)
>>> ?????? movl? %eax,?????? 0x04(%esp)
>>> ?????? call? _WHATEVER_THAT_IS
>>> ?????? testl %eax,?????? %eax
>>> ?????? je??? 0f
>>> ?? L00:jmp?? L00????????????????????? # loop forever
>>> ???? 0:hlt??????????????????????????? # CLI/STI?
>>> ?????? addl $0x08,?????? %esp
>>> ?????? ret
>>>
>>>
>>>
>>> As long as the code at 0x03E6 is unknown, it is impossible to tell
>>> anything. _H_Hat itself does nothing ... except wasting clocks and
>>> electrical power when the returned value is not zero.
>>>
>>> The only thing I can see in your debug trace is that you would run
>>> out of stack,
>
> If it doesn't get caught in the infinite loop at 603, likely less
> than one million before you hit the stack guard page.
>
It is only this portion of the excution trace of H_Hat that I am
referring to. The second call 000003e6 from the same machine address
[000005f2] without any control flow instructions inbetween indicates
infinite recursion. Many of the comp.theory people could not understand
this.
---[000005e6](01) 55 push ebp
---[000005e7](02) 8bec mov ebp,esp
---[000005e9](01) 51 push ecx
---[000005ea](03) 8b4508 mov eax,[ebp+08]
---[000005ed](01) 50 push eax
---[000005ee](03) 8b4d08 mov ecx,[ebp+08]
---[000005f1](01) 51 push ecx
---[000005f2](05) e8effdffff call 000003e6 --CALL [000003e6]
---[000005e6](01) 55 push ebp
---[000005e7](02) 8bec mov ebp,esp
---[000005e9](01) 51 push ecx
---[000005ea](03) 8b4508 mov eax,[ebp+08]
---[000005ed](01) 50 push eax
---[000005ee](03) 8b4d08 mov ecx,[ebp+08]
---[000005f1](01) 51 push ecx
---[000005f2](05) e8effdffff call 000003e6 --CALL
[000003e6]
>> That is the answer that I expected and confirms that I am correct.
>>
>> I am using the x86 code that was translated from C as the
>> machine description language of a Universal Turing Machine
>> equivalent. I wrote a whole x86utm operating system for
>> this purpose. This means that the assumption is infinite
>> memory and thus infinite stack.
>
> ??? I don't think any general-use compiler will emit a HLT
> instruction. No need, this is a priviliged ring0 instruction
> on modern x86 p-mode OSes.
>
> -- Robert
>
I use it in the x86 emulator to indicate the that system definitely
stopped executing at this point. Before I used this instruction the
emulator tended to try to execute garbage. When it popped the return
from main() it had no where to go.
--
Copyright 2020 Pete Olcott
"Great spirits have always encountered violent opposition from mediocre
minds." Einstein
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
