On Wed, 2017-10-11 at 06:10 -0700, Eric Dumazet wrote:
> On Tue, 2017-10-10 at 22:33 -0400, Manish Kurup wrote:
> > Using a spinlock in the VLAN action causes performance issues when the VLAN
> > action is used on multiple cores. Rewrote the VLAN action to use RCU read
> > locking for reads and updates instead.
> >
> > Signed-off-by: Manish Kurup <manish.ku...@verizon.com>
> > ---
> > include/net/tc_act/tc_vlan.h | 21 ++++++++-----
> > net/sched/act_vlan.c | 73
> > ++++++++++++++++++++++++++++++--------------
> > 2 files changed, 63 insertions(+), 31 deletions(-)
> >
> > diff --git a/include/net/tc_act/tc_vlan.h b/include/net/tc_act/tc_vlan.h
> > index c2090df..67fd355 100644
> > --- a/include/net/tc_act/tc_vlan.h
> > +++ b/include/net/tc_act/tc_vlan.h
> > @@ -13,12 +13,17 @@
> > #include <net/act_api.h>
> > #include <linux/tc_act/tc_vlan.h>
> >
> > +struct tcf_vlan_params {
> > + struct rcu_head rcu;
> > + int tcfv_action;
> > + u16 tcfv_push_vid;
> > + __be16 tcfv_push_proto;
> > + u8 tcfv_push_prio;
> > +};
> > +
> > struct tcf_vlan {
> > struct tc_action common;
> > - int tcfv_action;
> > - u16 tcfv_push_vid;
> > - __be16 tcfv_push_proto;
> > - u8 tcfv_push_prio;
> > + struct tcf_vlan_params __rcu *vlan_p;
> > };
> > #define to_vlan(a) ((struct tcf_vlan *)a)
> >
> > @@ -33,22 +38,22 @@ static inline bool is_tcf_vlan(const struct tc_action
> > *a)
> >
> > static inline u32 tcf_vlan_action(const struct tc_action *a)
> > {
> > - return to_vlan(a)->tcfv_action;
> > + return to_vlan(a)->vlan_p->tcfv_action;
>
> This is not proper RCU : ->vlan_p should be read once, and using
> rcu_dereference()
>
> So the caller should provide to this helper the 'p' pointer instead of
> 'a'
>
>
> CONFIG_SPARSE_RCU_POINTER=y
>
> and make C=2 would probably give you warnings about that.
>
BTW no need to change your .config after :
commit 41a2901e7d220875752a8c870e0b53288a578c20
Author: Paul E. McKenney <paul...@linux.vnet.ibm.com>
Date: Fri May 12 15:56:35 2017 -0700
rcu: Remove SPARSE_RCU_POINTER Kconfig option
The sparse-based checking for non-RCU accesses to RCU-protected pointers
has been around for a very long time, and it is now the only type of
sparse-based checking that is optional. This commit therefore makes
it unconditional.
Reported-by: Ingo Molnar <mi...@kernel.org>
Signed-off-by: Paul E. McKenney <paul...@linux.vnet.ibm.com>
Cc: Fengguang Wu <fengguang...@intel.com>
