Is IDT setup right ?

[macpacheco]

The original code from 64.asm seems to do this:
The first loop installs gates for INT 32 through INT 1. Skipping INT 0.
The 2nd loop installs gates for INT 21 -> exception_gate_00, INT 20 -> gate_01, ... INT 0 -> exception_gate_21. I think this is reversed.
The following code fixes the 2nd issue.
On the first one, should INT32 be reset ? No setting INT 0 handler at first gets rectified by 2nd loop.

	; Create exception gate stubs (Pure64 has already set the correct gate markers)
	xor edi, edi			; 64-bit IDT at linear address 0x0000000000000000
	mov ecx, 32
	mov rax, exception_gate		; A generic exception handler
make_exception_gate_stubs:
	call create_gate
	inc edi
	dec ecx
	jnz make_exception_gate_stubs

	; Set up the exception gates for all of the CPU exceptions
	xor edi, edi
	mov ecx, 21
	mov rax, exception_gate_21
make_exception_gates:
	call create_gate
	inc edi
	sub rax, 24			; Each exception gate is 24 bytes
	dec rcx
	jnz make_exception_gates

[IanSeyler]

I believe this is correct.

The first 32 entries (0-31) in the IDT are reserved for the CPU exceptions (of which 21 are used). Interrupts from 32 onwards are for hardware devices. I checked memory before and after creating the exception gate stubs and it set INT 0.

RDI is cleared to 0 so the first call to create_gate is for gate 0.

As a test try to do a divide by 0. Exception 0 should be triggered. You'll see the exception in the serial.log file in sys/

[macpacheco]

My mistake. I'm following the wrong variable. Its EDI that contains the gate number instead of rcx.

[IanSeyler]

No worries! It’s old code. ECX would make more sense.