Subject: Re: getting rid of NTOHS() in ip_input()
To: None <>
From: Jason R Thorpe <>
List: tech-net
Date: 08/13/2002 08:36:53
On Tue, Aug 13, 2002 at 04:02:58PM +0900, wrote:

 > 	i need to double-check reasssembly case.  other than that, it should
 > 	be safe.  i don't have any objection in having m_pullup() at the top of
 > 	ip_input() for safety, like:
 > 		if (m->m_flags & M_EXT)
 > 			m = m_pullup(m, sizeof(struct ip));

I think putting m_pullup() in ip_input() unconditionally would result in
an unnecessary performance penalty in the common case.

        -- Jason R. Thorpe <>