Re: [RT PATCH 2/2] net: add a lock around icmp_sk()

[Sebastian Andrzej Siewior]

On 2016-08-31 09:37:43 [-0700], Eric Dumazet wrote:
> > --- a/net/ipv4/icmp.c
> > +++ b/net/ipv4/icmp.c
> > @@ -216,12 +219,14 @@ static inline struct sock *icmp_xmit_lock(struct net *net)
> >  
> >  	local_bh_disable();
> >  
> > +	local_lock(icmp_sk_lock);
> 
> Deadlock alert ?
> Please read the comment few lines after, explaining why we have to use
> spin_trylock().
> Or maybe I should double check what is local_lock() in RT

On !RT local_lock() is preempt_disable().
On RT local_lock() is a per-CPU sleeping spinlock which can be taken
recursively by the owner.

On a "normal" ping reply we have
- icmp_reply()
  - icmp_xmit_lock()
    - local_lock()
  - icmp_push_reply()
 icmp_send()
    - local_lock()

so that works. I didn't manage to resolve a possible dst_link_failure()
path but I would assume it is like the above (where the owner takes the
same local_lock() multiple times).

Sebastian