[B.A.T.M.A.N.] [PATCH 1/2] batman-adv: convert orig_node->vlan_list to hlist

[B.A.T.M.A.N.] [PATCH 1/2] batman-adv: convert orig_node->vlan_list to hlist

[Marek Lindner]

Since the list's tail is never accessed using a double linked list head
wastes memory.

Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
---
 net/batman-adv/originator.c        | 6 +++---
 net/batman-adv/translation-table.c | 8 ++++----
 net/batman-adv/types.h             | 4 ++--
 3 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/net/batman-adv/originator.c b/net/batman-adv/originator.c
index 4500e3a..ce5e354 100644
--- a/net/batman-adv/originator.c
+++ b/net/batman-adv/originator.c
@@ -70,7 +70,7 @@ batadv_orig_node_vlan_get(struct batadv_orig_node *orig_node,
 	struct batadv_orig_node_vlan *vlan = NULL, *tmp;
 
 	rcu_read_lock();
-	list_for_each_entry_rcu(tmp, &orig_node->vlan_list, list) {
+	hlist_for_each_entry_rcu(tmp, &orig_node->vlan_list, list) {
 		if (tmp->vid != vid)
 			continue;
 
@@ -118,7 +118,7 @@ batadv_orig_node_vlan_new(struct batadv_orig_node *orig_node,
 	atomic_set(&vlan->refcount, 2);
 	vlan->vid = vid;
 
-	list_add_rcu(&vlan->list, &orig_node->vlan_list);
+	hlist_add_head_rcu(&vlan->list, &orig_node->vlan_list);
 
 out:
 	spin_unlock_bh(&orig_node->vlan_list_lock);
@@ -673,7 +673,7 @@ struct batadv_orig_node *batadv_orig_node_new(struct batadv_priv *bat_priv,
 		return NULL;
 
 	INIT_HLIST_HEAD(&orig_node->neigh_list);
-	INIT_LIST_HEAD(&orig_node->vlan_list);
+	INIT_HLIST_HEAD(&orig_node->vlan_list);
 	INIT_HLIST_HEAD(&orig_node->ifinfo_list);
 	spin_lock_init(&orig_node->bcast_seqno_lock);
 	spin_lock_init(&orig_node->neigh_list_lock);
diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c
index e96710c..ca0bcac 100644
--- a/net/batman-adv/translation-table.c
+++ b/net/batman-adv/translation-table.c
@@ -315,7 +315,7 @@ static void batadv_tt_global_size_mod(struct batadv_orig_node *orig_node,
 
 	if (atomic_add_return(v, &vlan->tt.num_entries) == 0) {
 		spin_lock_bh(&orig_node->vlan_list_lock);
-		list_del_rcu(&vlan->list);
+		hlist_del_rcu(&vlan->list);
 		spin_unlock_bh(&orig_node->vlan_list_lock);
 		batadv_orig_node_vlan_free_ref(vlan);
 	}
@@ -738,7 +738,7 @@ batadv_tt_prepare_tvlv_global_data(struct batadv_orig_node *orig_node,
 	u8 *tt_change_ptr;
 
 	rcu_read_lock();
-	list_for_each_entry_rcu(vlan, &orig_node->vlan_list, list) {
+	hlist_for_each_entry_rcu(vlan, &orig_node->vlan_list, list) {
 		num_vlan++;
 		num_entries += atomic_read(&vlan->tt.num_entries);
 	}
@@ -764,7 +764,7 @@ batadv_tt_prepare_tvlv_global_data(struct batadv_orig_node *orig_node,
 	(*tt_data)->num_vlan = htons(num_vlan);
 
 	tt_vlan = (struct batadv_tvlv_tt_vlan_data *)(*tt_data + 1);
-	list_for_each_entry_rcu(vlan, &orig_node->vlan_list, list) {
+	hlist_for_each_entry_rcu(vlan, &orig_node->vlan_list, list) {
 		tt_vlan->vid = htons(vlan->vid);
 		tt_vlan->crc = htonl(vlan->tt.crc);
 
@@ -2463,7 +2463,7 @@ static void batadv_tt_global_update_crc(struct batadv_priv *bat_priv,
 
 	/* recompute the global CRC for each VLAN */
 	rcu_read_lock();
-	list_for_each_entry_rcu(vlan, &orig_node->vlan_list, list) {
+	hlist_for_each_entry_rcu(vlan, &orig_node->vlan_list, list) {
 		/* if orig_node is a backbone node for this VLAN, don't compute
 		 * the CRC as we ignore all the global entries over it
 		 */
diff --git a/net/batman-adv/types.h b/net/batman-adv/types.h
index da4c738..e298332 100644
--- a/net/batman-adv/types.h
+++ b/net/batman-adv/types.h
@@ -190,7 +190,7 @@ struct batadv_vlan_tt {
 struct batadv_orig_node_vlan {
 	unsigned short vid;
 	struct batadv_vlan_tt tt;
-	struct list_head list;
+	struct hlist_node list;
 	atomic_t refcount;
 	struct rcu_head rcu;
 };
@@ -302,7 +302,7 @@ struct batadv_orig_node {
 	spinlock_t out_coding_list_lock; /* Protects out_coding_list */
 #endif
 	struct batadv_frag_table_entry fragments[BATADV_FRAG_BUFFER_COUNT];
-	struct list_head vlan_list;
+	struct hlist_head vlan_list;
 	spinlock_t vlan_list_lock; /* protects vlan_list */
 	struct batadv_orig_bat_iv bat_iv;
 };
-- 
2.1.4

[B.A.T.M.A.N.] [PATCH 2/2] batman-adv: prevent potential hlist double deletion

[Marek Lindner]

The hlist_del_rcu() call in batadv_tt_global_size_mod() does not check
if the element still is part of the list prior to deletion. The atomic
list counter should prevent the worst but converting to
hlist_del_init_rcu() ensures the element can't be deleted more than
once.

Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
---
 net/batman-adv/translation-table.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c
index ca0bcac..e29c9e1 100644
--- a/net/batman-adv/translation-table.c
+++ b/net/batman-adv/translation-table.c
@@ -315,7 +315,7 @@ static void batadv_tt_global_size_mod(struct batadv_orig_node *orig_node,
 
 	if (atomic_add_return(v, &vlan->tt.num_entries) == 0) {
 		spin_lock_bh(&orig_node->vlan_list_lock);
-		hlist_del_rcu(&vlan->list);
+		hlist_del_init_rcu(&vlan->list);
 		spin_unlock_bh(&orig_node->vlan_list_lock);
 		batadv_orig_node_vlan_free_ref(vlan);
 	}
-- 
2.1.4

Re: [B.A.T.M.A.N.] [PATCH 1/2] batman-adv: convert orig_node->vlan_list to hlist

[Antonio Quartulli]

[-- Attachment #1: Type: text/plain, Size: 289 bytes --]



On 21/06/15 18:30, Marek Lindner wrote:
> Since the list's tail is never accessed using a double linked list head
> wastes memory.
> 
> Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>

Acked-by: Antonio Quartulli <antonio@meshcoding.com>


-- 
Antonio Quartulli


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [B.A.T.M.A.N.] [PATCH 2/2] batman-adv: prevent potential hlist double deletion

[Antonio Quartulli]

[-- Attachment #1: Type: text/plain, Size: 1368 bytes --]



On 21/06/15 18:30, Marek Lindner wrote:
> The hlist_del_rcu() call in batadv_tt_global_size_mod() does not check
> if the element still is part of the list prior to deletion. The atomic
> list counter should prevent the worst but converting to
> hlist_del_init_rcu() ensures the element can't be deleted more than
> once.
> 
> Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>

Acked-by: Antonio Quartulli <antonio@meshcoding.com>

However, as discussed offline after Sven's suggestion in ticket #217,
the entire if-loop still needs to be protected with the vlan_list_lock
and not just its body.

Cheers,

> ---
>  net/batman-adv/translation-table.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c
> index ca0bcac..e29c9e1 100644
> --- a/net/batman-adv/translation-table.c
> +++ b/net/batman-adv/translation-table.c
> @@ -315,7 +315,7 @@ static void batadv_tt_global_size_mod(struct batadv_orig_node *orig_node,
>  
>  	if (atomic_add_return(v, &vlan->tt.num_entries) == 0) {
>  		spin_lock_bh(&orig_node->vlan_list_lock);
> -		hlist_del_rcu(&vlan->list);
> +		hlist_del_init_rcu(&vlan->list);
>  		spin_unlock_bh(&orig_node->vlan_list_lock);
>  		batadv_orig_node_vlan_free_ref(vlan);
>  	}
> 

-- 
Antonio Quartulli


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [B.A.T.M.A.N.] [PATCH 1/2] batman-adv: convert orig_node->vlan_list to hlist

[Marek Lindner]

[-- Attachment #1: Type: text/plain, Size: 375 bytes --]

On Monday, June 22, 2015 18:37:52 Antonio Quartulli wrote:
> On 21/06/15 18:30, Marek Lindner wrote:
> > Since the list's tail is never accessed using a double linked list head
> > wastes memory.
> >
> > 
> >
> > Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
> 
> Acked-by: Antonio Quartulli <antonio@meshcoding.com>

Applied in revision 8d85d13.

Regards,
Marek

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

Re: [B.A.T.M.A.N.] [PATCH 2/2] batman-adv: prevent potential hlist double deletion

[Marek Lindner]

[-- Attachment #1: Type: text/plain, Size: 781 bytes --]

On Monday, June 22, 2015 18:44:20 Antonio Quartulli wrote:
> On 21/06/15 18:30, Marek Lindner wrote:
> > The hlist_del_rcu() call in batadv_tt_global_size_mod() does not check
> > if the element still is part of the list prior to deletion. The atomic
> > list counter should prevent the worst but converting to
> > hlist_del_init_rcu() ensures the element can't be deleted more than
> > once.
> >
> > Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
> 
> Acked-by: Antonio Quartulli <antonio@meshcoding.com>

Applied in revision 4729a33.

 
> However, as discussed offline after Sven's suggestion in ticket #217,
> the entire if-loop still needs to be protected with the vlan_list_lock
> and not just its body.

I'll take care of that in a separate patch.

Regards,
Marek

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 473 bytes --]