On Thu, 2 Jan 2014, Ding Tianhong wrote: > On 2014/1/2 18:26, Ding Tianhong wrote: > > On 2014/1/2 17:14, Julia Lawall wrote: > >> On Thu, 2 Jan 2014, Ding Tianhong wrote: > >> > >>> On 2014/1/2 16:38, Julia Lawall wrote: > >>>> On Thu, 2 Jan 2014, Ding Tianhong wrote: > >>>> > >>>>> On 2014/1/2 15:39, Julia Lawall wrote: > >>>>>> Are the casts needed > >>>>> > >>>>> > >>>>> Yes, otherwise the warming will report: > >>>>> > >>>>> /net-next/drivers/net/bonding/bond_3ad.c:427: warning: passing argument > >>>>> 1 of ‘ether_addr_equal_64bits’ from incompatible pointer type > >>>> > >>>> Is it necessary for this driver to use a different type from everyone > >>>> else? > >>>> > >>>> julia > >>>> > >>> Did you mean the MAC_ADDRESS_EQUAL is excess? > >>> I did not remove it because the codes no need to be changed more and it > >>> looks that didn't take any negative effect. > >> > >> No, I was wondering about the mac_addr type, defined in bond_3ad.h. Other > >> code just has the array inlined into the containing structure. > >> > >> julia > >> > > I reviewed the struct mac_addr again, and feel that even it looks not > comfortable, but > make the lacp struct more meaning for 3ad, what do you think about it, I > think no need > to revert them to u8.
Personally, when I see things that are different, I start wondering about why. So if there is no reason for it to be different, I would prefer that it is the same. Certainly, a mac_addr type is more meaningful than just an array with size ETH_ALEN, or worse an array with size 6. But I am not sure that it is practical to introduce that type everywhere. In any case, it is not a big issue. julia
