General Protection Fault

[matthias]

Hi everyone,

When I enable interrupts, I get a general protection fault (0x0D). GDT, IDT code is ok. I used the same code as my c++ kernel but, now I get a GPF :S

Code: Select all

init_gdt(); // goes well.. debuging didn't point any bugs
init_idt(); // the same for this one.. 

asm("sti"); // after this a GPF happens :@(6)
for(;;); // go and do infinitly nothing :p

I really have no idea why this happened.. A GPF when I'm just looping :S
I hope someone can help.

Here is also some source:
http://www.telebyte.nl/~vdvlies/source.rar

Thanks in advance,

Matthias

[Osbios]

GDT, IDT code is ok. I used the same code as my c++ kernel

Oo ?

I really have no idea why this happened.. A GPF when I'm just looping :S

hmm... you dont know timer ints?
http://www.clipx.net/ng/hardware/ng256d3.php

And btw, i cant C or C++.

[matthias]

Sorry, but this has nothing to do with timer ints.. This should work since it worked for me before.. I only stripped the code out of my classes and converted it to normal C code.

[xenos]

I just looked at your code. What happens, when the timer interrupt fires? There doesn't seem to be an interrupt handler for that case, resulting in a GPF...

[earlz]

hmm i bet thats why my os gets a gpf after enabling interrupts and then going into an infinite loop

you also need to remap the pic because by default it starts at 8 which is also reserved for exceptions so it would be like everytime your timer rings it has an 8th exception

[matthias]

you also need to remap the pic because by default it starts at 8 which is also reserved for exceptions so it would be like everytime your timer rings it has an 8th exception

If you've read my source code carefully I already did that

[matthias]

I just looked at your code. What happens, when the timer interrupt fires? There doesn't seem to be an interrupt handler for that case, resulting in a GPF...

Though there is.. The interrupt is handled by the default handler int_unknown() which does only an iret (just lookup in my idt_init() function).. Shouldn't give a problem. Worked before.

Code: Select all

void int_unknown()
{
	asm("iret");
}

[matthias]

After some debugging I found out that the GPF occured after the asm("iret"); So IDT is ok.. Maybe it's of any use.. I wonder what causes this :S

[xenos]

Have you checked whether the compiler creates a stack frame for int_unknown? If so, the raw asm("iret"); causes a return to the pushed EBP register instead of the return address, resulting in a GPF.

[matthias]

Have you checked whether the compiler creates a stack frame for int_unknown? If so, the raw asm("iret"); causes a return to the pushed EBP register instead of the return address, resulting in a GPF.

That could be the cause of my problem.. Thanks.. I'll put it in a .asm file.. and test if it works

[bubach]

I would suggest that you insert an EOI before the "iret"....

[matthias]

EOI wasn't needed to solve the problem.. I've put the function in a assembly file and assebled it with nasm.. linked it in my kernel.. and it works... But Now I've got another question (not a problem :p).. Is there an option in GCC to disable the creation of a stack-frame? (Was it the -fomit-frame-pointer option ??)

[earlz]

in my kernel i had to pop 3 dwords to trash(edx) for it to iret to the interrupt pushed stuff

also you mght not get a gpf for it but you should send an end of interrupt signal to the pic

such as here is my timer interrupt function

Code: Select all

#define irqm_return __asm("pop %edx");__asm("pop %edx");__asm("pop %edx");outportb(0x20, 0x20);__asm("iret")
void irq0(){ /*overflow causes invalid opcode error, although that dont quite make sense*/
	/*timer*/
	if (timer_ticks>0xFFFFFFF5){timer_ticks=0;}
	timer_ticks++;
	irqm_return;
}

as i stated, for some reason UNDER BOCHS an overflow causes an invalid opcode error although i got no error with microsoft vpc and i got no error on my real pc