Jump distance is ignored if the jump location is same

[tomsons26]

I have a asm written function where there's alignment in between jumps

extracted relevant asm code

.686P
.mmx
.model	flat, C

		.code;


; no idea how to get this to appear so fake it
_align MACRO count
	REPT (count / 8)
		;lea esp, [esp+0] but zero is optimized to a byte..
		db 08Dh, 0A4h, 024h, 00h, 00h, 00h, 00h
	ENDM
ENDM

_aligna MACRO reg
	add reg,0
ENDM

_alignl MACRO reg
	lea reg,[reg]
ENDM

_alignm MACRO reg
	mov reg,reg
ENDM

Adjust_Color_1 proc C uses ebx ecx edx esi edi

    punpckhwd mm2, mm1
    punpckhwd mm2, mm2
    jmp     ??mmx_loop

    _align 32
    _align 16

??mmx_loop:
    add     edi, 2
Adjust_Color_1 endp

Adjust_Color_2 proc C uses ebx ecx edx esi edi

    punpckhwd mm2, mm1
    punpckhwd mm2, mm2
    jmp     ??mmx_loop

    _align 32
    _align 8
    db 08Dh, 064h, 024h, 00h

??mmx_loop:
    add     edi, 2
Adjust_Color_2 endp

end

This should still be flagged as something

assembled obj
objdiffjmpissue.zip

[tomsons26]

#233 partially related to this as this also has a explicit vs implicit displacement

[tomsons26]

Perhaps this could be handled by showing 100% but not green(a thing objdiff already does for numeric mismatches only) and showing the jump displacement on the side?
Mockup

[encounter]

Yeah, this is a tricky one. The primary issue is that the instruction sizes are different but emit the same disassembly. objdiff only caring about the jump target is intended behavior, it helps with partially matched functions.

I think what needs to happen is to mark an instruction as mismatching if the instruction size is different.

[encounter]

I think that idea would also fix #233