segment problem

[assembler01]

This is my 16 bit kernel code:

Code: Select all

section .text
bits 16
org 0x0
_start:
cli
mov ax, 0x0
mov ss, ax
mov sp, 0x0ffff
sti
cld
mov ax, 0x0200 ; the segment my bootloader loads my kernel
mov ds, ax
mov es, ax
mov fs, ax
mov gs, ax
test: ; this tests to see if the bootloader has actually booted my kernel
mov ah, 0x0e
mov al, 'O'
xor bh, bh
int 0x10
main:
call clean_all ; clears the screen and registers
mov si, .msg
call print
jmp .halt
.msg: db 'Testing, testing ...', 0
.halt:
cli
hlt
jmp .halt
section .data
%include "screen.asm"
%include "clean.asm"

When I run it in qemu it boots up perfectly (it shows the 'O' char) but after that, nothing happends. I guess it is a segmentation problem but the setup is like in the mikeos kernel.

[Yoda]

Why org 0?
BIOS loads boot sector to 7C00h.
Also, moving 0FFFFh into SP makes stack not word-aligned.

[assembler01]

This is not my bootloader, this is the (test) kernel. I have tested the bootloader loading the mikeos kernel and it worked perfectly.

[Yoda]

How bootloader jumps to kernel? I.e. what is in CS register?

[qw]

When I run it in qemu it boots up perfectly (it shows the 'O' char) but after that, nothing happends.

Perhaps it's because you disable interrupts before the HLT. I know this prevends Bochs from updating the screen, maybe the same counts for Qemu.

[Yoda]

The cli/hlt is his final sequence. That's ok.
I think, he found that didn't properly initialize CS by far jump, that's why he didn't respond to last my post.

[xenos]

Perhaps it's because you disable interrupts before the HLT. I know this prevends Bochs from updating the screen, maybe the same counts for Qemu.

I'm not quite sure about Bochs, but I know for sure that it prevents updating the screen on Qemu.

[bubach]

Still, how is that relevant? By that point the message is already printed, and it's not exactly uncommon to end your code with cli, hlt.

[VolTeK]

Set your segments before your jump to your kernel code.

Why org 0?

Because as you can see, its his kernel code. Not his bootloader code.

[qw]

Still, how is that relevant? By that point the message is already printed

No, it is not. It is printed when Bochs/Qemu refreshes the screen, which may be a long time (relatively speaking) after return from INT 10H.

[Yoda]

Because as you can see, its his kernel code. Not his bootloader code.

Ask others to repeat it too. Did you read this topic? I did.

Hey, guys, topicstarter probably solved his problem and left this topic. What is the reason to plow the sand after him?

[qw]

Yoda, a forum is a way to exchange ideas, not just solve some OP's problem. Bubach obviously was not aware of the effect of CLI/HLT on certain emulators, he is now.