netdev
[Top] [All Lists]

RE: obvious ndisc.c bug

To: "'David S. Miller'" <davem@xxxxxxxxxx>
Subject: RE: obvious ndisc.c bug
From: Ra Chen <Ra@xxxxxxxxxxxxxxxxxxx>
Date: Fri, 20 Apr 2001 10:08:20 +0800
Cc: "'netdev@xxxxxxxxxxx'" <netdev@xxxxxxxxxxx>
Sender: owner-netdev@xxxxxxxxxxx

> From: David S. Miller [mailto:davem@xxxxxxxxxx]
> Ra Chen writes:
>  > net/ipv6/ndisc.c in kernel 2.4.3
>  > line 1076:
>  >
>  > if ((ipv6_addr_type(saddr)&IPV6_ADDR_MULTICAST) &&
>  >
>  > "saddr" should be "daddr".
>
> This code looks correct to me, what exactly makes you think it should
> be testing "daddr"?  Similar code above for the solication cases also
> checks "saddr" for being multicast.

I think a multicast packet is a packet with multicast destination address. Am I wrong?
It's impossible for a source address to be multicast, right?
Where's the "similar code"?  As I noticed line 1015 and 1043 both check "daddr" for being multicast.

Ra Chen

<Prev in Thread] Current Thread [Next in Thread>