Page 1 of 1
HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 10:50 am
by Jlouro
I was reading this
ATA read/write sectors to read IDE through ports.
I made this assembly code:
Code: Select all
;-----------------------------------Disk Start-----------------------------------------;
global _readd
global _writed
;--------------------------------Read Disk Start---------------------------------------;
_readd:
;push eax
;push ebx
;push ecx
;push edx
push ebp
mov ebp, esp
mov eax, [ebp+8] ;first parameter
mov ecx, eax
cmp ecx, 256
jg short _readsector_fail
jne short _readsector
xor ecx, ecx
_readsector:
mov ebx, ecx
mov eax, [ebp+12] ;second parameter
cmp eax, 0
je _readsector_fail
push eax ;sector number
mov dx, 0x1f2
mov al, cl
out dx, al
inc dx ;dx = 0x1f3
pop eax
out dx, al
inc dx ;dx = 0x1f4
shr eax, 8
out dx, al
inc dx ;dx = 0x1f5
shr eax, 8
out dx, al
inc dx ;dx = 0x1f6
shr eax, 8
and al, 00001111b
or al, 01000000b
out dx, al
inc dx ;dx = 0x1f7
mov al, 0x20
out dx, al
mov ecx, 4
_readsector_wait:
in al, dx
test al, 0x80 ; BSY
jne _readsector_retry
test al, 0x08 ; DRQ
jne _readsector_ready
_readsector_retry:
dec ecx
jg _readsector_wait
_readsector_nextsector:
in al, dx
test al, 0x80
jne _readsector_nextsector
test al, 0x21
jne _readsector_fail
_readsector_ready:
sub dx, 7 ;0x1f0
mov ecx, 256
rep insw
add dx, 7
in al, dx ;wait 400 ns
in al, dx
in al, dx
in al, dx
dec ebx
jg _readsector_nextsector
;mov eax, edi
;cmp eax, 0x0ffffffff
;jz _readsector_fail
mov [ebp-8], edi ;Second local variable
leave
ret
_readsector_fail:
xor eax, eax
leave
ret
;-----------------------Read Disk End-------------------------------------;
It is always printing a comma. Is it legal to do that mov [ebp-8], edi ?
In this same file I have writedisk and I mov esi, [ebp-4] (which is the first local variable).
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:01 am
by iansjack
I find it very difficult to follow your code as there are no useful comments and no indication of what the parameters and variables represent. A couple of questions:
1. What do you mean by "printing a comma". Unless I missed something I see no calls to print routines.
2. You don't seem to ever use EDI or assign anything to it and then, out of nowhere, you assign its value to a local variable. Why? Especially as you then immediately leave the function.
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:03 am
by BASICFreak
Can you please rephrase your question?
It is always printing a comma. Is it legal to do that mov [ebp-8], edi ?
I see nowhere in that code segment that has anything to do with printing or a comma.
And yes; mov [ebp - 8], edi is fully legal.
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:05 am
by Jlouro
iansjack wrote:I find it very difficult to follow your code as there are no useful comments and no indication of what the parameters and variables represent. A couple of questions:
1. What do you mean by "printing a comma". Unless I missed something I see no calls to print routines.
2. You don't seem to ever use EDI or assign anything to it and then, out of nowhere, you assign its value to a local variable. Why? Especially as you then immediately leave the function.
I'm not used to work with EDI. I have a kernel.c with PrintString. You want me to post here my kernel.c code?
When I rep insw It should retrieve the value to edi right? My ebp-8 is a local variable in c
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:09 am
by Jlouro
BASICFreak wrote:Can you please rephrase your question?
It is always printing a comma. Is it legal to do that mov [ebp-8], edi ?
I see nowhere in that code segment that has anything to do with printing or a comma.
And yes; mov [ebp - 8], edi is fully legal.
Code: Select all
extern int _readd(int sector_count, int nmrsector);
extern int _writed(int sector_count, int nmrsector);
void kmain()
{
int msg = 104;
int recv;
int sector = 1;
int read = 1;
int writeerror = _writed(read, sector);
if(writeerror == 0) PrintString("Error Writing\n");
int error = _readd(read, sector);
if(error == 0) PrintString("ERROR");
PrintChar((char)recv);
PrintChar('\n');
PrintString("READ!");
readStr();
}
That's my kernel.c file. I'm probably printing that wrong.
ebp-8 will be recv right?
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:14 am
by BASICFreak
INSW - reads word from port DX into address ES:(E)DI
You have not set EDI, insw will not set it, only increase it by 2 for each word you read from port DX. (without setting (E)DI you can be overwriting anything)
http://x86.renejeschke.de/html/file_mod ... d_141.html
ebp-8 will be recv right?
No, if you want to pass recv - pass it as an argument do not rely on the stack to have the value - it's probably in the .BSS section.
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:24 am
by Jlouro
BASICFreak wrote:INSW - reads word from port DX into address ES:(E)DI
You have not set EDI, insw will not set it, only increase it by 2 for each word you read from port DX. (without setting (E)DI you can be overwriting anything)
http://x86.renejeschke.de/html/file_mod ... d_141.html
ebp-8 will be recv right?
No, if you want to pass recv - pass it as an argument do not rely on the stack to have the value - it's probably in the .BSS section.
so is it better to do: _readd(int, int, int) and then make mov edi, [ebp+16] ?
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:50 am
by BASICFreak
so is it better to do: _readd(int, int, int) and then make mov edi, [ebp+16] ?
The way a standard C call works, any modification to the values passed to the function will be lost. The only way to keep the value is to pass it as a pointer (int*) and modify it in memory.
Here is your call frame.
e.g.:
Code: Select all
_readd(int*, int, int)
_readd(&recv, read, sector)
mov ebx, [ebp + 12] ; I may be wrong with the 12 here
mov [ebx], edi
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 11:57 am
by iansjack
Jlouro wrote:BASICFreak wrote:
ebp-8 will be recv right?
Wrong!
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 12:00 pm
by Jlouro
BASICFreak wrote:so is it better to do: _readd(int, int, int) and then make mov edi, [ebp+16] ?
The way a standard C call works, any modification to the values passed to the function will be lost. The only way to keep the value is to pass it as a pointer (int*) and modify it in memory.
Here is your call frame.
e.g.:
Code: Select all
_readd(int*, int, int)
_readd(&recv, read, sector)
mov ebx, [ebp + 12] ; I may be wrong with the 12 here
mov [ebx], edi
One more question:
Code: Select all
_asm_function(int a, int b, int c)
mov ebx, [ebp+12]
In that code am I moving my variable b to ebx ? I read that in C arguments are passed left-to-right and [ebp+4] is the return.
So [ebp+8] will be int a;
[ebp+12] will be int b;
[ebp+16] will be int c. Right?
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 12:08 pm
by BASICFreak
It's pushed Right to Left
So your pushes onto the stack are
push C (+20)
push B (+16)
push A (+12)
push ReturnAddress (+8)
push EBP (+4)
Re: HardDrive - PMode x86 assembly
Posted: Sat Feb 20, 2016 12:10 pm
by iansjack
You know, there is an easy way to test this sort of question. Run your code in a debugger and look at what actually happens.