Re: [PATCH 1/5] TCP infrastructure split out

[Baruch Even <baruch@ev-en.org>]

Stephen Hemminger wrote:

 > @@ -1356,7 +1301,7 @@
 >  static inline void __tcp_enter_cwr(struct tcp_sock *tp)
 >  {
 >  	tp->undo_marker = 0;
 > -	tp->snd_ssthresh = tcp_recalc_ssthresh(tp);
 > +	tp->snd_ssthresh = tp->ca_proto->ssthresh(tp);

 > +void tcp_ca_register(struct tcp_ca_type *ca)
 > +{
 > +	BUG_ON(tcp_ca_find(ca->name));
 > +
 > +	spin_lock_irq(&tcp_ca_list_lock);
 > +	list_add_tail_rcu(&ca->list, &tcp_ca_list);
 > +	spin_unlock_irq(&tcp_ca_list_lock);
 > +
 > +	printk(KERN_INFO "TCP %s registered\n", ca->name);
 > +}

Since we have some calls to callbacks unprotected, shouldn't we check 
them on tcp_ca_register to make sure they are in fact available so that 
we don't needlessly crash?

Baruch