svn commit: r271174 - head/sys/sys

[Robert N. M. Watson]

On 7 Sep 2014, at 11:23, Gleb Smirnoff <glebius at FreeBSD.org> wrote:

> R> Modified: head/sys/sys/mbuf.h
> R> ==============================================================================
> R> --- head/sys/sys/mbuf.h	Fri Sep  5 16:40:47 2014	(r271173)
> R> +++ head/sys/sys/mbuf.h	Fri Sep  5 16:46:28 2014	(r271174)
> R> @@ -672,7 +672,7 @@ m_clget(struct mbuf *m, int how)
> R>  {
> R>  
> R>  	if (m->m_flags & M_EXT)
> R> -		printf("%s: %p mbuf already has cluster\n", __func__, m);
> R> +		printf("%s: %p mbuf already has external storage\n", __func__, m);
> R>  	m->m_ext.ext_buf = (char *)NULL;
> R>  	uma_zalloc_arg(zone_clust, m, how);
> R>  	/*
> R> @@ -698,7 +698,7 @@ m_cljget(struct mbuf *m, int how, int si
> R>  	uma_zone_t zone;
> R>  
> R>  	if (m && m->m_flags & M_EXT)
> R> -		printf("%s: %p mbuf already has cluster\n", __func__, m);
> R> +		printf("%s: %p mbuf already has external storage\n", __func__, m);
> R>  	if (m != NULL)
> R>  		m->m_ext.ext_buf = NULL;
> 
> Shouldn't these be KASSERTs? I'm pretty sure that Andre put it as printf(9)
> because he was not sure it is correct. But since that time we never seen
> these prints in the wild, so they can be safely converted to KASSERTs.

Hi Gleb:

I agree that they should be KASSERT()s -- almost all if (x) panic(..) should be -- but was working offline at the time and so didn't have a chance to spelunk the mailing lists or GNATS/Bugzilla to see if I could find references to the panics so left the current form as is. I am currently on travel but have a TODO to do some spelunking when I get back. I spotted a number of other similar constructs in the mbuf code, which I'm currently working through fairly closely looking at the variable-size mbuf changes, which depend on a number of cleanups. (The ones I've made thus far are cosmetic, but less purely cosmetic ones will follow soon.)

Robert