[solved] CPUID trouble

Index > OS Construction > [solved] CPUID trouble

Goto page 1, 2 Next

Author

Thread

Coddy41

Joined: 18 Jan 2009
Posts: 384
Location: Ohio, USA

Coddy41 29 Jan 2010, 01:33

OK, let us all assume were in p-mode. I have been trying to get the CPUID according to
wikipedia all I have to do is make eax = 0 use the 'cpuid' instruction and it stores the
return stuff into ebx, ecx, and edx. I then have to store all this somewhere. But I cant
get it to work.

Here is my CPUID code.

Code:

  Mov   ESI, msg
  Call   PrintString
  Call    Get_CPUID
  Mov    ESI, _CPUID
  Call    PrintString
  jmp     $
Get_CPUID:
   Push  EAX  ; Save regs we need to use later on.
   Push  ECX
   Push  EDX
   xor   eax, eax   ; get vendor ID
   cpuid          ; CPUID will now be saved into EBX, ECX, and EDX. So we need to
                  ; store that, should not be to hard. :-/
   Mov   ESI, _CPUID  ; This is were we store the CPU ID for now 
   mov   DWORD [ESI], EBX
   mov   DWORD [ESI+4], ECX
   mov   DWORD [ESI+8], EDX
; OK, so that is done! Lets restore what regs we saved and blow this joint!
   Pop   EDX
   Pop   ECX
   Pop   EDX
   ret
;;;;;;;;;;;;;;;;;;;Data;;;;;;;;;;;;;;;;;;;
msg  db "CPUID: ",0
_CPUID db 0,0,0,0,0,0,0,0,0,0,0,0

I can't seem to find what I am doing wrong, many other examples seem to do it
close to the same way. Pointers would be appropriated Very Happy

PS: I am just assuming user has a Pentium+ for now.

_________________
Want hosting for free for your asm project? You can PM me. (*.fasm4u.net)

Last edited by Coddy41 on 30 Jan 2010, 17:01; edited 1 time in total

29 Jan 2010, 01:33

dosin

Joined: 24 Aug 2007
Posts: 337

dosin 29 Jan 2010, 04:35

this a prints cpuname and small vend name in pmode:
hope this helps!

Code:

pusha
mov   eax,0       ; eax=0

cpuid

    
mov   [cpuname+ 12],ebx
mov   [cpuname+ 16],edx
mov   [cpuname+ 20],ecx
mov   [smallvendor],ecx
 
popa
  
mov   esi,cpuname
call print_string
  
mov   esi,spacer
call print_string
   
mov   esi,smallvendor
call print_string

jmp $

cpuname: db 'CPU VENDOR:             ',0
spacer: db " ",0
smallvendor dd ?

Last edited by dosin on 29 Jan 2010, 14:46; edited 1 time in total

29 Jan 2010, 04:35

roboman

Joined: 03 Dec 2006
Posts: 122
Location: USA

roboman 29 Jan 2010, 05:53

Don't know much about cpuid, but does your PrintString choke if it's not 0 terminated?

29 Jan 2010, 05:53

sinsi

Joined: 10 Aug 2007
Posts: 789
Location: Adelaide

sinsi 29 Jan 2010, 06:05

Watch your pushes and pops too

Code:

   Push  EAX  ; Save regs we need to use later on. 
   Push  ECX 
   Push  EDX 
;...
   Pop   EDX 
   Pop   ECX 
   Pop   EDX

29 Jan 2010, 06:05

tom tobias

Joined: 09 Sep 2003
Posts: 1320
Location: usa

tom tobias 29 Jan 2010, 12:16

hmm. EDX, EAX, hmm. Now where have we had this discussion before????

Yup, that's right, a single letter change, leading to an error, which is difficult to identify....difficult to debug....hahaha...

Proof is in the pudding.....

Smile

29 Jan 2010, 12:16

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20340
Location: In your JS exploiting you and your system

revolution 29 Jan 2010, 12:41

When I first saw that the first person I thought of was tom tobias.

tom, thanks for not letting me down. Smile

So what should be do about this "problem"? Clearly it is not acceptable that one single letter change is allowed to make a program crash (regardless of whether it is in and xor or a push). Should we rename all the registers and opcodes to ensure a minimum possible Hamming distance?

29 Jan 2010, 12:41

edfed

Joined: 20 Feb 2006
Posts: 4335
Location: Now

edfed 29 Jan 2010, 13:47

the single letter change is like the single DNA change.
it gives unpredictable results.

a way to push:pop regs is preferable
do them on a single line, with the 4 general registers.(not only 3).

Code:

push eax ebx excx edx
pop edx ecx ebx eax

then, you just have to compare the lines, the first in forward order, the second in reverse order.

and if you are very lazy, you can do it with a macro

Code:

macro pushg32{
push eax ebx ecx edx
}
macro popg32{
pop edx ecx ebx eax
}

push eax ebx ecx edx is better because if you want to add a new functionality in your code, you can do it with ALL registers.
after, you do the optimisation with a reduced pushing/poping.

and all with lowercase chars is better because faster to write. Smile

29 Jan 2010, 13:47

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20340
Location: In your JS exploiting you and your system

revolution 29 Jan 2010, 13:50

edfed: Have you ever seen the 'proc' macro?

29 Jan 2010, 13:50

edfed

Joined: 20 Feb 2006
Posts: 4335
Location: Now

edfed 29 Jan 2010, 13:52

i know 'proc', but i don't use it.
never, i don't understand it... Sad

like 'struct'...

29 Jan 2010, 13:52

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20340
Location: In your JS exploiting you and your system

revolution 29 Jan 2010, 13:55

edfed wrote:

i know 'proc', but i don't use it.

Then that is why you have to type everything twice.

29 Jan 2010, 13:55

edfed

Joined: 20 Feb 2006
Posts: 4335
Location: Now

edfed 29 Jan 2010, 14:08

revolution wrote:

edfed wrote:
i know 'proc', but i don't use it.
Then that is why you have to type everything twice.

cool, i'll give it a try one day. Smile

but it is an abstraction layer, no?

_________________
Smile

fool gitlab fasmstuff
gitlab foolstuff

29 Jan 2010, 14:08

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20340
Location: In your JS exploiting you and your system

revolution 29 Jan 2010, 14:53

edfed wrote:

but it is an abstraction layer, no?

Yes. Much the same as your pushg32/popg32 macros above.

29 Jan 2010, 14:53

Coddy41

Joined: 18 Jan 2009
Posts: 384
Location: Ohio, USA

Coddy41 29 Jan 2010, 17:05

roboman wrote:

Don't know much about cpuid, but does your PrintString choke if it's not 0 terminated?

Nah, It would probably just print a bunch of ascii characters to the screen until a 0 comes along.

Hmm, I fixed that push pop problem, but it still does not seem to work Confused

I have decided to do a re-write with it based from dosin's example.

EDIT: Funny, even dosin's example would not work in my OS, I am wondering if it is the
GDT I am using, I am going to wright my own and see. I'll post back if it works Confused

_________________
Want hosting for free for your asm project? You can PM me. (*.fasm4u.net)

29 Jan 2010, 17:05

bitshifter

Joined: 04 Dec 2007
Posts: 796
Location: Massachusetts, USA

bitshifter 29 Jan 2010, 17:21

Also i made one for MM testing... (still not complete)
http://board.flatassembler.net/topic.php?t=10708

29 Jan 2010, 17:21

Coddy41

Joined: 18 Jan 2009
Posts: 384
Location: Ohio, USA

Coddy41 29 Jan 2010, 17:28

Wow I did a search on the fasm board on 'CPUID' before making this thread and I do not
think I saw that one =/ Thanks I'll study it as I wanted to test for MMX Smile

_________________
Want hosting for free for your asm project? You can PM me. (*.fasm4u.net)

29 Jan 2010, 17:28

dosin

Joined: 24 Aug 2007
Posts: 337

dosin 29 Jan 2010, 19:30

What you could do - is make a dex os app that will display it..
since dex os is in pmode and have all the functions you need ...print string ect..

once you have it to print what you want - and working right.. transfer to your os.. then if its still not working.. look at the functions your using - your print text... gdt... pmode code... ect.

29 Jan 2010, 19:30

bitshifter

Joined: 04 Dec 2007
Posts: 796
Location: Massachusetts, USA

bitshifter 29 Jan 2010, 19:51

Dex already have that program written (CPUID.DEX)
Maybe he would be kind enough to share the sources Smile

29 Jan 2010, 19:51

Coddy41

Joined: 18 Jan 2009
Posts: 384
Location: Ohio, USA

Coddy41 29 Jan 2010, 21:09

@dosin: It works in DexOS wich is weird, Im going to check the rest of my code
and see if that works.

@bitshifter: Atually, the CPUID source is posted on the DexOS forums, do a search
of "CPUID not working" The code has a bug, but if you know what you are doing there
is a post that tells you the problem Wink

_________________
Want hosting for free for your asm project? You can PM me. (*.fasm4u.net)

29 Jan 2010, 21:09

dosin

Joined: 24 Aug 2007
Posts: 337

dosin 29 Jan 2010, 23:40

if you have a print char function you can try that and see if it has the vender name stored - just print one char at a time..

if it works than there may be something in your printstring...

if still nothing .. it could be something with your GDT segment set up...
your data segs.. or something..

Code:

call get_cpuid_info

mov al,byte[cpuname+ 12]
call print_char
mov al,byte[cpuname+ 13]
call print_char
mov al,byte[cpuname+ 14]
call print_char
ect...

29 Jan 2010, 23:40

Coddy41

Joined: 18 Jan 2009
Posts: 384
Location: Ohio, USA

Coddy41 30 Jan 2010, 00:28

My printstring works fine after dex pointed out the problem (see here http://dex.7.forumer.com/viewtopic.php?p=5812#5812)

But it does not hurt to try... I got my same result Sad

_________________
Want hosting for free for your asm project? You can PM me. (*.fasm4u.net)

30 Jan 2010, 00:28

Goto page 1, 2 Next

< Last Thread | Next Thread >

Forum Rules:

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum