Re: [PATCH] ipv6: Keep index within tab_unreach[]

[Brian Haley <brian.haley@hp.com>]

David Miller wrote:
> From: Brian Haley <brian.haley@hp.com>
> Date: Wed, 17 Jun 2009 22:10:08 -0400
> 
>> Roel Kluin wrote:
>>> @@ -923,7 +923,7 @@ int icmpv6_err_convert(int type, int code, int *err)
>>>  	switch (type) {
>>>  	case ICMPV6_DEST_UNREACH:
>>>  		fatal = 1;
>>> -		if (code <= ICMPV6_PORT_UNREACH) {
>>> +		if (code <= ICMPV6_PORT_UNREACH && code >= 0) {
>>>  			*err  = tab_unreach[code].err;
>>>  			fatal = tab_unreach[code].fatal;
>>>  		}
>> The code value in the ICMPv6 header is a u8, so should always be positive, right?
>> It doesn't hurt I guess though.
> 
> True.  Probably best to pass this thing in as a 'u8', rathern than as
> an 'int'.

"type" is a u8 too, but both seem to be passed as int's everywhere (rawv6,
tcp, sctp and dccp), and in the err_handler() func pointer in the inet6_protocol
struct.  I can try and cook something up in the next few days that changes
it all to u8's.

-Brian