Recent full disclosure post - Local DOS

[Lawrence Stewart]

On 01/29/11 11:30, Christian Peron wrote:
> On Fri, Jan 28, 2011 at 02:27:18PM -0500, John Baldwin wrote:
> [..]
>> ===================================================================
>> --- tcp_usrreq.c	(revision 218018)
>> +++ tcp_usrreq.c	(working copy)
>> @@ -1330,7 +1330,8 @@ tcp_ctloutput(struct socket *so, struct sockopt *s
>>  				tp->t_flags |= TF_NOPUSH;
>>  			else {
>>  				tp->t_flags &= ~TF_NOPUSH;
>> -				error = tcp_output(tp);
>> +				if (TCPS_HAVEESTABLISHED(tp->t_state))
>> +					error = tcp_output(tp);
>>  			}
>>  			INP_WUNLOCK(inp);
>>  			break;
> 
> I was thinking of correcting it the same way.. I might even do something
> like:
> 
> 	else {
> 		if (tp->t_flags & TF_NOPUSH) {
> 			tp->t_flags &= ~TF_NOPUSH;
> 			if (TCPS_HAVEESTABLISHED(tp->t_state))
> 				error = tcp_output(tp);
> 		}
> 	}
> 
> By default, this mask is not set.. so un-setting it and calling tcp_output() 
> if it was not already set seems wasteful

Apologies for tuning in late, but FWIW I concur and think the above
patch is appropriate.

Cheers,
Lawrence