Re: [PATCH v1 net-next] ila: avoid genlmsg_reply when not ila_map found

[Simon Horman]

On Thu, May 23, 2024 at 09:44:15AM +0800, Lin Ma wrote:
> Hi Simon.
> 
> > 
> > Hi Lin Ma,
> > 
> > The lines immediately above those covered by this patch are as follows:
> > 
> > 	ret = -ESRCH;
> > 	ila = ila_lookup_by_params(&xp, ilan);
> > 	if (ila) {
> > 		ret = ila_dump_info(ila,
> > 
> > > @@ -483,6 +483,8 @@ int ila_xlat_nl_cmd_get_mapping(struct sk_buff *skb, struct genl_info *info)
> > >  				    info->snd_portid,
> > >  				    info->snd_seq, 0, msg,
> > >  				    info->genlhdr->cmd);
> > > +	} else {
> > > +		ret = -EINVAL;
> > >  	}
> > >  
> > >  	rcu_read_unlock();
> > 
> > And the lines following, up to the end of the function, are:
> > 
> > 	if (ret < 0)
> > 		goto out_free;
> > 
> > 	return genlmsg_reply(msg, info);
> > 
> > out_free:
> > 	nlmsg_free(msg);
> > 	return ret;
> > 
> > By my reading, without your patch, if ila is not found (NULL)
> > then ret will be -ESRCH, and genlmsg_reply will not be called.
> > 
> > I feel that I am missing something here.
> 
> Oh my bad, it seems this bug was already fixed by the
> commit 693aa2c0d9b6 ("ila: do not generate empty messages
> in ila_xlat_nl_cmd_get_mapping()") last year.
> And my dated kernel does not apply that one.
> 
> Thanks for reminding me of this false alarm.

Thanks for checking.
I think we can retire this patch.