Very lost on botched interrupts causing triple fault

[MichaelPetch]

The fact it is ignoring -T *is* a serious problem. As a result The linker script (Linker.ld) you are specifying is being ignored. What happens when you remove `-z`. That `-z` is bogus but the `-T` after it is not.

You can believe whatever you want about whatever options you are using but I can tell you this - Until you ultimately fix your linker options you are going to tear you hair out trying to fix that 0x804xxxxxx address causing all kinds of problem. Your triple fault right now is because you are telling the processor to go to a bogus memory address that doesn't actually have a service routine at it. I will make you a bet that you have a bunch of 0x804..... addresses in your kernel where mine looks like this:

Code: Select all

objdump -x _build/iso/System/Kernel/NovaVita.kernel

_build/iso/System/Kernel/NovaVita.kernel:     file format elf32-i386
_build/iso/System/Kernel/NovaVita.kernel
architecture: i386, flags 0x00000102:
EXEC_P, D_PAGED
start address 0x00101288

Program Header:
    LOAD off    0x00001000 vaddr 0x00100000 paddr 0x00100000 align 2**12
         filesz 0x00001000 memsz 0x00001000 flags r--
    LOAD off    0x00002000 vaddr 0x00101000 paddr 0x00101000 align 2**12
         filesz 0x00004004 memsz 0x0000d000 flags rwx
   STACK off    0x00000000 vaddr 0x00000000 paddr 0x00000000 align 2**4
         filesz 0x00000000 memsz 0x00000000 flags rwx

Sections:
Idx Name          Size      VMA       LMA       File off  Algn
  0 .multiboot    00001000  00100000  00100000  00001000  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  1 .text         00002000  00101000  00101000  00002000  2**4
                  CONTENTS, ALLOC, LOAD, READONLY, CODE
  2 .rodata       00001000  00103000  00103000  00004000  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  3 .eh_frame     000000fc  00104000  00104000  00005000  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  4 .data         00000f08  001040fc  001040fc  000050fc  2**2
                  CONTENTS, ALLOC, LOAD, DATA
  5 .gdt          00000ffc  00105004  00105004  00006004  2**0
                  ALLOC
  6 .bss          00008000  00106000  00106000  00006004  2**12
                  ALLOC
  7 .comment      0000003d  00000000  00000000  00006004  2**0
                  CONTENTS, READONLY
  8 .gnu_debuglink 0000001c  00000000  00000000  00006044  2**2
                  CONTENTS, READONLY
SYMBOL TABLE:
no symbols

You'll note I don't have addresses that start with 0x804.... and this is what happens every time a timer interrupt fires is:

Code: Select all

0: v=08 e=0000 i=0 cpl=0 IP=0008:001012a2 pc=001012a2 SP=0010:0010a000 env->regs[R_EAX]=001032f0

When I press a key I get an interrupt and it doesn't triple fault and it looks like this:

Code: Select all

v=09 e=0000 i=0 cpl=0 IP=0008:001012a2 pc=001012a2 SP=0010:0010a000 env->regs[R_EAX]=001032f0

Once you remap the PIC(s) and create proper entries in your IDT and service routines for them you are going to still encounter triple faults. You can ignore the problem I am telling you about and you will be back here asking why it is still faulting on you.

[austanss]

Alright, so I wrote the PIC controller configuration function. As it is C++ code, do I extern it as C and call it from assembly? Or would I ideally call it in loadGDT/loadIDT?

[Octocontrabass]

It doesn't matter too much when you call it, as long as you call it before you set the interrupt flag. (You probably can't call it before your global constructors if it's C++ code.)

[austanss]

The fact it is ignoring -T *is* a serious problem. As a result The linker script (Linker.ld) you are specifying is being ignored. What happens when you remove `-z`. That `-z` is bogus but the `-T` after it is not.

You can believe whatever you want about whatever options you are using but I can tell you this - Until you ultimately fix your linker options you are going to tear you hair out trying to fix that 0x804xxxxxx address causing all kinds of problem. Your triple fault right now is because you are telling the processor to go to a bogus memory address that doesn't actually have a service routine at it. I will make you a bet that you have a bunch of 0x804..... addresses in your kernel where mine looks like this:

Code: Select all

objdump -x _build/iso/System/Kernel/NovaVita.kernel

_build/iso/System/Kernel/NovaVita.kernel:     file format elf32-i386
_build/iso/System/Kernel/NovaVita.kernel
architecture: i386, flags 0x00000112:
EXEC_P, HAS_SYMS, D_PAGED
start address 0x00101278

Program Header:
    PHDR off    0x00000034 vaddr 0x000ff034 paddr 0x000ff034 align 2**2
         filesz 0x000000e0 memsz 0x000000e0 flags r--
  INTERP off    0x000050fc vaddr 0x001040fc paddr 0x001040fc align 2**0
         filesz 0x00000013 memsz 0x00000013 flags r--
    LOAD off    0x00000000 vaddr 0x000ff000 paddr 0x000ff000 align 2**12
         filesz 0x00002000 memsz 0x00002000 flags r--
    LOAD off    0x00002000 vaddr 0x00101000 paddr 0x00101000 align 2**12
         filesz 0x00004094 memsz 0x0000d000 flags rwx
 DYNAMIC off    0x00006004 vaddr 0x00105004 paddr 0x00105004 align 2**2
         filesz 0x00000090 memsz 0x00000090 flags rw-
EH_FRAME off    0x0000513c vaddr 0x0010413c paddr 0x0010413c align 2**2
         filesz 0x00000044 memsz 0x00000044 flags r--
   STACK off    0x00000000 vaddr 0x00000000 paddr 0x00000000 align 2**4
         filesz 0x00000000 memsz 0x00000000 flags rwx

Dynamic Section:
  GNU_HASH             0x00104124
  STRTAB               0x00104120
  SYMTAB               0x00104110
  STRSZ                0x00000001
  SYMENT               0x00000010
  DEBUG                0x00000000
  REL                  0x00104180
  RELSZ                0x00000a40
  RELENT               0x00000008
  TEXTREL              0x00000000
  FLAGS                0x0000000c
  FLAGS_1              0x08000001
  RELCOUNT             0x00000148

Sections:
Idx Name          Size      VMA       LMA       File off  Algn
  0 .multiboot    00001000  00100000  00100000  00001000  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  1 .text         00002000  00101000  00101000  00002000  2**4
                  CONTENTS, ALLOC, LOAD, READONLY, CODE
  2 .rodata       00001000  00103000  00103000  00004000  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  3 .eh_frame     000000fc  00104000  00104000  00005000  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  4 .interp       00000013  001040fc  001040fc  000050fc  2**0
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  5 .dynsym       00000010  00104110  00104110  00005110  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  6 .dynstr       00000001  00104120  00104120  00005120  2**0
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  7 .gnu.hash     00000018  00104124  00104124  00005124  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  8 .eh_frame_hdr 00000044  0010413c  0010413c  0000513c  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
  9 .rel.dyn      00000a40  00104180  00104180  00005180  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, DATA
 10 .data         00000444  00104bc0  00104bc0  00005bc0  2**2
                  CONTENTS, ALLOC, LOAD, DATA
 11 .dynamic      00000090  00105004  00105004  00006004  2**2
                  CONTENTS, ALLOC, LOAD, DATA
 12 .gdt          00000f6c  00105094  00105094  00006094  2**0
                  ALLOC
 13 .bss          00008000  00106000  00106000  00006094  2**12
                  ALLOC
 14 .comment      0000003d  00000000  00000000  00006094  2**0
                  CONTENTS, READONLY
 15 .gnu_debuglink 0000001c  00000000  00000000  000060d4  2**2
                  CONTENTS, READONLY
SYMBOL TABLE:
no symbols

You'll note I don't have addresses that start with 0x804.... and this is what happens every time a timer interrupt fires is:

Code: Select all

0: v=08 e=0000 i=0 cpl=0 IP=0008:001012a2 pc=001012a2 SP=0010:0010a000 env->regs[R_EAX]=001032f0

When I press a key I get an interrupt and it doesn't triple fault and it looks like this:

Code: Select all

v=09 e=0000 i=0 cpl=0 IP=0008:001012a2 pc=001012a2 SP=0010:0010a000 env->regs[R_EAX]=001032f0

Once you remap the PIC(s) and create proper entries in ISR.asm you are going to still encounter triple faults. You can ignore the problem I am telling you about and you will be back here asking why it is still faulting on you.

You Are A Legend. I Love You. Thank You So Much.
The triple faults are gone, however the interrupts don't work. Most likely because of my handler logic.

My hate for you used to be agonizing, but now you have exposed me to my flaws, and your righteousness.

[austanss]

It doesn't matter too much when you call it, as long as you call it before you set the interrupt flag. (You probably can't call it before your global constructors if it's C++ code.)

Sorry for a dumb question, but what are the parameters I am supposed to pass? The parameters are offset1 and offset2, both of type int.

Got the function from https://wiki.osdev.org/8259_PIC#Program ... e_8259_PIC

[Octocontrabass]

The page you copied the code from explains what those parameters are and gives an example for how you might set them.

[austanss]

The page you copied the code from explains what those parameters are and gives an example for how you might set them.

In the hacker land, these people are referred to as script kiddies, the most annoying and intolerable known to man.

[Octocontrabass]

There are also links to other pages like this one that might explain it more clearly.

[austanss]

There are also links to other pages like this one that might explain it more clearly.

I still don't know what I'm supposed to set those parameters as...

[Octocontrabass]

Code: Select all

/*
arguments:
	offset1 - vector offset for master PIC
		vectors on the master become offset1..offset1+7
	offset2 - same for slave PIC: offset2..offset2+7
*/

You're supposed to set each one to the vector offset for each PIC.

Pick the vector offset based on which interrupt vectors you want to use for each PIC.

It's common to use 0x20 and 0x28 since those are the first available vectors after the ones reserved for exceptions, but it's your OS so you can choose whatever you want.

[austanss]

Code: Select all

/*
arguments:
	offset1 - vector offset for master PIC
		vectors on the master become offset1..offset1+7
	offset2 - same for slave PIC: offset2..offset2+7
*/

You're supposed to set each one to the vector offset for each PIC.

Pick the vector offset based on which interrupt vectors you want to use for each PIC.

It's common to use 0x20 and 0x28 since those are the first available vectors after the ones reserved for exceptions, but it's your OS so you can choose whatever you want.

I did all this, fine and dandy, I configure PIC and all. I wrote a simple IRQ handler that prints a keycode if possible. Not sure if it worked, but either way I wrote the function to print the interrupt number code. When attempting to trigger an interrupt (e.g. pressing the keyboard) I get no response. Not sure about the keyboard, as I'm using QEMU, and a USB keyboard, and my driver only supports PS/2, and I'm unsure if QEMU (from command line, no established VM) uses PS/2 emulation by default. Either way, nothing is happening from the IRQ handler. I've made sure that the IRQ handler is externed as C.

Relevant: the repo - files "GDT.cpp" and "ISR.asm"

[Octocontrabass]

Did you set the interrupt flag?

[austanss]

Did you set the interrupt flag?

No, I suppose I should do that... uh, how do I do that?? I thought it happened automatically, but in hindsight I didn't really see an attributes or anything on the handler function.

[Octocontrabass]

The STI instruction sets the interrupt flag. If the interrupt flag is clear, the CPU will not receive any interrupts from external hardware.

[austanss]

The STI instruction sets the interrupt flag. If the interrupt flag is clear, the CPU will not receive any interrupts from external hardware.

oh i uh. i already have that instruction.

from Boot.S

Code: Select all

_start:
        /* stack */
        mov $stack_top, %esp

        cli

        call loadGDT

        mov %cr0, %eax
        or 1, %al
        mov %eax, %cr0

        call loadIDT

        call configurePIC

        jmp $0x8, $protectedModeMain

protectedModeMain:

        sti

        /* here we call kernel_main */
        call kernel_main