Receiving only one Keyboard IRQ using IOAPIC

[stdcall]

Hi.
I'm implementing 8042 keyboard driver and I only manage to get single IRQ when pressing a key, all other key presses / releases won't trigger an interrupt.

I do acknowledge the IRQ to the local APIC by writing 0 to APIC_EOI_REGISTER.
and in the keyboard IRQ handler I read a single byte from the PS2 data port.

I didn't see any explicit EOI I need to write to in the PS2/Keyboard interface.
Did I miss something else ?

PS. The keyboard works correctly in polling mode.

[Brendan]

Hi,

I didn't see any explicit EOI I need to write to in the PS2/Keyboard interface.
Did I miss something else ?

You get an IRQ when the PS/2 controller's 1-byte buffer goes from "empty" to "full". If you don't remove the byte from the buffer, then the buffer stays full and can't go from "empty" to "full" again.


Cheers,

Brendan

[stdcall]

But I do read a single byte. and I even check if there's more and there isn't.

[BrightLight]

Try running in Bochs. Does it show any log messages? There are some sources on the Internet saying that the local APIC EOI register is 0xA0. This is wrong; the EOI register is 0xB0.
BTW, did you try the PIT before the keyboard?

[stdcall]

OK. took me some time to configure bochs. but here it is.
do you see something suspicious?

Code: Select all

00128488000d[KBD   ] gen_scancode(): BX_KEY_W pressed
00128488000d[KBD   ] keyboard: gen_scancode with scancode_translate cleared
00128488000d[KBD   ] gen_scancode(): writing raw 1d 
00128488000d[KBD   ] kbd_enQ(0x1d)
00128488000d[KBD   ] kbd_enQ: putting scancode 0x1d in internal buffer 
00128488000d[KBD   ] activating timer...
00128488000d[KBD   ] service_keyboard: key in internal buffer waiting
00128489000d[IOAPIC] set_irq_level(): INTIN1: level=1
00128489000d[IOAPIC] IOAPIC: servicing
00128489000d[APIC0 ] Deliver lowest priority of fixed interrupt vector 21
00128489000d[APIC0 ] trigger interrupt vector=0x21
00128489000d[APIC0 ] triggered vector 0x21
00128489000d[APIC0 ] lapic(0): not delivering int 0x21 because int 0x21 is in service
00128489000d[IOAPIC] service_ioapic(): INTIN2 is masked 
00128489036i[CPU0  ] RDMSR: Read 00000000:fee00800 from MSR_APICBASE
00128489056d[APIC0 ] LAPIC write 0x00000000 to register 0x0b00 
00128489056e[APIC0 ] APIC write: register ffffffff not implemented
00128489101d[IOAPIC] set_irq_level(): INTIN1: level=0
00128489101d[KBD   ] READ(60) = 1d
00128489696d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489756d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489781d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489821d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489841d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489866d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489891d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489916d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489941d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489961d[IOAPIC] set_irq_level(): INTIN4: level=0
00128489986d[IOAPIC] set_irq_level(): INTIN4: level=0
00128490000d[KBD   ] service_keyboard(): no keys waiting

[stdcall]

issue resolved. it was bad LAPIC configuration. I wonder how I even received the first IRQ.

[SpyderTL]

Can you be more specific on the problem and the solution? Switching from the PIC to APIC is at the top of my list of TO-DOs.

Thanks.

[crunch]

Can you be more specific on the problem and the solution? Switching from the PIC to APIC is at the top of my list of TO-DOs.

Thanks.

Ditto that. I want to get my LAPIC code finished this week. Got SMP initialization working though!

[BrightLight]

Can you be more specific on the problem and the solution? Switching from the PIC to APIC

Ditto that. I want to get my LAPIC code finished this week. Got SMP initialization working though!

This is enough for my local APIC code to work.

Code: Select all

	; configure the local APIC
	mov rax, 0
	mov rcx, 0x80
	call lapic_write_register

	mov rax, 0x1FF
	mov rcx, 0xF0
	call lapic_write_register	; enable spurious IRQ at INT 0xFF

	mov rcx, 0xE0
	mov eax, 0xFFFFFFFF
	call lapic_write_register