disk read error

[scales of justice]

hello,

i tried to load my some data with the int 0x13,
but it doesn't work, the error code is 0x10
i looked in a list:
0x10 means "read error"

what can be the problem?
"read error" is helping a lot

[gaf]

Hello,
from what I know it's possible that the floppy just didn't manage to spin-up in the given amount of time. You should try to read for a few times before giving up..

regards,
gaf

[scales of justice]

I tried that too, but it didn't make any difference
but maybe I've to wait longer...

[gaf]

According to this list of error-code, 0x10 means "uncorrectable CRC or ECC error on read". A read error is code 0x04 and a timeout would be 0x80.

What parameters are you using to call int 0x13 ? Maybe you're trying to access a non-existing sector and your BIOS just returns the wrong error-code.

regards,
gaf

[scales of justice]

you're right, I made a mistake in my function, which should read the error code
the right error code is 0x01, he means "invalid function in AH or invalid parameter",

that's strange, the program runs in bochs without problems,
only my old ami-bios makes problems

maybe yousee the mistake:

Code: Select all

readdisk:
push bp
mov bp, sp
sub sp, 0x0006
push ax
push bx
push cx
push dx
push es

mov ax, [bp+0x04]
mov bx, 0x0024
xor dx, dx
div bx
mov [bp-0x02], ax
mov ax, dx
push ax
mov bx, 0x0012
xor dx, dx
div bx
mov [bp-0x04], ax
pop ax
xor dx, dx
div bx  
mov ax, dx
inc ax
mov [bp-0x06], ax

mov ax, [bp+0x06]
mov es, ax
xor bx, bx
mov ah, 0x02
mov al, 0x01
mov ch, [bp-0x02]
mov cl, [bp-0x06]
mov dh, [bp-0x04]
mov dl, 0x00
int 0x13

pop es
pop dx
pop cx
pop bx
pop ax
mov sp, bp
pop bp
ret

the first parameter contains the the logical sector and he second one the segment

But it can't be a mathematical problem in the function, in bochs he reads exactly the correct sectors

[gaf]

Code: Select all

mov ax, [bp+0x04] ; ax = logical sector
mov bx, 0x0024
xor dx, dx
div bx
mov [bp-0x02], ax ; bp-2 = sector/0x24
mov ax, dx
push ax           ; push [sector%24]
mov bx, 0x0012    ; sectors per track
xor dx, dx
div bx
mov [bp-0x04], ax ; bp-4 = (sector%0x24)/0x12
pop ax
xor dx, dx
div bx
mov ax, dx
inc ax
mov [bp-0x06], ax ; bp-6 = ((sector%24)%0x12) + 1

mov ax, [bp+0x06]
mov es, ax        ; es = segment                 (es:bx buffer)
xor bx, bx        ; bx = 0
mov ah, 0x02      ; ah = 2                       (function)
mov al, 0x01      ; al = 1                       (number of sectors)
mov ch, [bp-0x02] ; ch = sector/0x24             (lower 9bits of track number)
mov cl, [bp-0x06] ; cl = ((sector%0x24)%0x12) +1 (sector number)
mov dh, [bp-0x04] ; dh = (sector%0x24)/0x12      (head number)
mov dl, 0x00      ; dl = 0                       (drive number)
int 0x13

Hmm, I've gone through your code, and either my assembler is much worse than I thought it was, or you mixed up some of the facts about floppy disk geometry.
The first thing is that there is 18 sectors per track and not 36, except if you're using a 2.88 MB floppy, which however seems rather unlikely to me. This means that you have to change the value in line 2, so that CH will eventually be sector/0x12. To get the sector number (CL) all you have to do is to take the remainer of the division. The sector is the offset in the current track and there should be no need to do any additionaly divsions as you do in lines 6-10. To get the head number, all you have to do is to divide the logical sector number by the number of sectors per head (1440).

I also suggest that you add one to the logical sector number at the very beginning of your calculations, as it will influence all other results.

cheers,
gaf

[scales of justice]

this time you're not right:

36 is correct, at first you must divide with the sectors per Track and then with the heads, I combined these two steps

look here:
http://www.osdever.net/tutorials/chs_lba.php?the_id=87


but I found the real problem now:

if there are more than 10 bytes on the stack int 13h doesn't work
but I can't imagine why o.O


EDIT:
no it works
I changed the position of the stack
maybe there was anything of my bios

[gaf]

Ok, had a look at the tutorial and it seems as if you were right about the heads. It's actually been some time that I did this for the last time and I'm sorry if I just added to the confusion..

While you can basically forget about my whole explanation, I still think that the comments I added to your code are essentially correct. In that case however the formulas given in the tutorial only agree with your algorithm for the sectors, but not for the head- and track numbers. It might be that your code works for some LBA addresses, but I'm quite sure that it will fail once you're trying to load something significantly large - such as a kernel.

regards,
gaf