Subject: Re: multiple copies of arp_lock_try()
To: Atsushi Onoe <firstname.lastname@example.org>
From: Thor Lancelot Simon <email@example.com>
Date: 06/24/2002 02:43:51
On Mon, Jun 24, 2002 at 01:47:28PM +0900, Atsushi Onoe wrote:
> > another question - why do we need if_ieee1394arp.c? i took a diff
> > between if_arp.c and i saw almost no difference...
> Hmm. I can't recall the reason but it seems simply because 1394ARP is
> defined differently from normal ARP. It doesn't have target hardware
> address, and the length of hardware address is defined as 16 rather
> than 8 (IEEE1394_ADDR_LEN).
> But as you see, we designed 1394 interface with address length 16 which is
> the same format as 1394ARP. The difference in code is minimal as a result.
> So we may merge it into if_arp.c, though my knowledge about 1394 is almost
Incidentally, is there any reason you can see to not support a larger MTU
as an interface option on 1394 interfaces? I know 1394 ARP doesn't use
enough of the available bits to express the real hardware MTU, but it might
be nice to be able to just use the extra bits if all hosts involved were
(I do not understand why the choice was made to artifically constrain the
MTU in 1934 ARP -- do you?)