You can also do it by hooking two ports up in reverse. like p1.0 to p3.7, p1.1 to p3.6 and so on until p1.7 to p3.0. Then

mov p1,a
mov a,p3

uhhh, well, it's more of a joke than it's meant to be a practical answer ;-)
Michae's look-up table approach is probably the best.
If you want to do it without a LUT, then

mov c,acc.0
mov b.7,c
mov c,acc.1
mov b.6,c
mov c,acc.2
mov b.5,c
mov c,acc.3
mov b.4,c
mov c,acc.4
mov b.3,c
mov c,acc.5
mov b.2,c
mov c,acc.6
mov b.1,c
mov c,acc.7
mov b.0,c
mov a,b

That's slower, but it takes less space in code memory.