Re: [PATCH 2/2] ipvs: Guard access of HK_TYPE_KTHREAD cpumask with RCU

[Waiman Long]

On 3/26/26 4:32 AM, Julian Anastasov wrote:
> 	Hello,
>
> On Tue, 24 Mar 2026, Waiman Long wrote:
>
> > The ip_vs_ctl.c file and the associated ip_vs.h file are the only places
> > in the kernel where HK_TYPE_KTHREAD cpumask is being retrieved and used.
> > Now that HK_TYPE_KTHREAD/HK_TYPE_DOMAIN cpumask can be changed at run
> > time. We need to use RCU to guard access to this cpumask to avoid a
> > potential UAF problem as the returned cpumask may be freed before it
> > is being used.
> >
> > Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
> > ---
> >   include/net/ip_vs.h            | 20 ++++++++++++++++----
> >   net/netfilter/ipvs/ip_vs_ctl.c | 13 ++++++++-----
> >   2 files changed, 24 insertions(+), 9 deletions(-)
> >
> > diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
> > index 29a36709e7f3..17c85a575ef4 100644
> > --- a/include/net/ip_vs.h
> > +++ b/include/net/ip_vs.h
> > @@ -1155,7 +1155,7 @@ static inline int sysctl_run_estimation(struct netns_ipvs *ipvs)
> >   	return ipvs->sysctl_run_estimation;
> >   }
> >   
> > -static inline const struct cpumask *sysctl_est_cpulist(struct netns_ipvs *ipvs)
> > +static inline const struct cpumask *__sysctl_est_cpulist(struct netns_ipvs *ipvs)
> >   {
> >   	if (ipvs->est_cpulist_valid)
> >   		return ipvs->sysctl_est_cpulist;
> > @@ -1273,7 +1273,7 @@ static inline int sysctl_run_estimation(struct netns_ipvs *ipvs)
> >   	return 1;
> >   }
> >   
> > -static inline const struct cpumask *sysctl_est_cpulist(struct netns_ipvs *ipvs)
> > +static inline const struct cpumask *__sysctl_est_cpulist(struct netns_ipvs *ipvs)
> >   {
> >   	return housekeeping_cpumask(HK_TYPE_KTHREAD);
> >   }
> > @@ -1290,6 +1290,18 @@ static inline int sysctl_est_nice(struct netns_ipvs *ipvs)
> >   
> >   #endif
> >   
> 	May be there is little fuzz here, due to the recent
> changes in the nf-next tree. If this is a bugfix due to the
> missing RCU protection, may be you should add Fixes line too
> and use the nf tree. Probably, there will be fuzz/collisions with
> the changes in the nf-next tree...
Thanks for the suggestion, I will rebase the patches on top of the 
nf-next tree.
> > +static inline bool sysctl_est_cpulist_empty(struct netns_ipvs *ipvs)
> > +{
> > +	guard(rcu)();
> > +	return cpumask_empty(__sysctl_est_cpulist(ipvs));
> > +}
> > +
> > +static inline unsigned int sysctl_est_cpulist_weight(struct netns_ipvs *ipvs)
> > +{
> > +	guard(rcu)();
> > +	return cpumask_weight(__sysctl_est_cpulist(ipvs));
> > +}
> > +
> >   /* IPVS core functions
> >    * (from ip_vs_core.c)
> >    */
> > @@ -1604,7 +1616,7 @@ static inline void ip_vs_est_stopped_recalc(struct netns_ipvs *ipvs)
> >   	/* Stop tasks while cpulist is empty or if disabled with flag */
> >   	ipvs->est_stopped = !sysctl_run_estimation(ipvs) ||
> >   			    (ipvs->est_cpulist_valid &&
> > -			     cpumask_empty(sysctl_est_cpulist(ipvs)));
> > +			     sysctl_est_cpulist_empty(ipvs));
> >   #endif
> >   }
> >   
> > @@ -1620,7 +1632,7 @@ static inline bool ip_vs_est_stopped(struct netns_ipvs *ipvs)
> >   static inline int ip_vs_est_max_threads(struct netns_ipvs *ipvs)
> >   {
> >   	unsigned int limit = IPVS_EST_CPU_KTHREADS *
> > -			     cpumask_weight(sysctl_est_cpulist(ipvs));
> > +			     sysctl_est_cpulist_weight(ipvs);
> >   
> >   	return max(1U, limit);
> >   }
> > diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
> > index 35642de2a0fe..f38a2e2a9dc5 100644
> > --- a/net/netfilter/ipvs/ip_vs_ctl.c
> > +++ b/net/netfilter/ipvs/ip_vs_ctl.c
> > @@ -1973,11 +1973,14 @@ static int ipvs_proc_est_cpumask_get(const struct ctl_table *table,
> >   
> >   	mutex_lock(&ipvs->est_mutex);
> >   
> > -	if (ipvs->est_cpulist_valid)
> > -		mask = *valp;
> > -	else
> > -		mask = (struct cpumask *)housekeeping_cpumask(HK_TYPE_KTHREAD);
> > -	ret = scnprintf(buffer, size, "%*pbl\n", cpumask_pr_args(mask));
> > +	/* HK_TYPE_KTHREAD cpumask needs RCU protection */
> 	Can we switch IPVS to use HK_TYPE_DOMAIN? The initial
> intention was to follow the code in kthread.c. Then you can
> reconsider if HK_TYPE_KTHREAD should be alias to HK_TYPE_DOMAIN,
> may be not if there are no other users...

Yes, I can certainly switch to use HK_TYPE_DOMAIN instead. The reason I 
keep HK_TYPE_KTHREAD is that it may not be obvious to others that 
kthread is now following the HK_TYPE_DOMAIN cpumask, 
keeping  HK_TYPE_KTHREAD but making it an alias can make that clear.


> > +	scoped_guard(rcu) {
> > +		if (ipvs->est_cpulist_valid)
> > +			mask = *valp;
> > +		else
> > +			mask = (struct cpumask *)housekeeping_cpumask(HK_TYPE_KTHREAD);
> > +		ret = scnprintf(buffer, size, "%*pbl\n", cpumask_pr_args(mask));
> > +	}
> >   
> >   	mutex_unlock(&ipvs->est_mutex);
> >   
> > --
> > 2.53.0
> Regards
>
> --
> Julian Anastasov <ja@xxxxxx>