[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <aAnXfryT1sYcE7-m@shredder>
Date: Thu, 24 Apr 2025 09:17:34 +0300
From: Ido Schimmel <idosch@...dia.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: netdev@...r.kernel.org, davem@...emloft.net, pabeni@...hat.com,
edumazet@...gle.com, andrew+netdev@...n.ch, razor@...ckwall.org,
petrm@...dia.com, roopa@...dia.com
Subject: Re: [PATCH net] vxlan: vnifilter: Fix unlocked deletion of default
FDB entry
On Wed, Apr 23, 2025 at 02:29:21PM -0700, Jakub Kicinski wrote:
> On Wed, 23 Apr 2025 17:51:31 +0300 Ido Schimmel wrote:
> > I'm sorry, but I only noticed this issue after the recent VXLAN patches
> > were applied to net-next. There will be a conflict when merging net into
> > net-next, but resolution is trivial. Reference:
> > https://github.com/idosch/linux/commit/ed95370ec89cccbf784d5ef5ea4b6fb6fa0daf47.patch
>
> Thanks! I guess this shouldn't happen often but FWIW for conflict-less
> build breakage a patch on top of the merge would be more convenient
> than the net-next version of the patch. Like this:
No problem, but note that "hash_index" needs to be removed as well, so
this would be the diff:
diff --git a/drivers/net/vxlan/vxlan_vnifilter.c b/drivers/net/vxlan/vxlan_vnifilter.c
index 81d088c2f8dc..186d0660669a 100644
--- a/drivers/net/vxlan/vxlan_vnifilter.c
+++ b/drivers/net/vxlan/vxlan_vnifilter.c
@@ -625,10 +625,7 @@ static void vxlan_vni_delete_group(struct vxlan_dev *vxlan,
*/
if (!vxlan_addr_any(&vninode->remote_ip) ||
!vxlan_addr_any(&dst->remote_ip)) {
- u32 hash_index = fdb_head_index(vxlan, all_zeros_mac,
- vninode->vni);
-
- spin_lock_bh(&vxlan->hash_lock[hash_index]);
+ spin_lock_bh(&vxlan->hash_lock);
__vxlan_fdb_delete(vxlan, all_zeros_mac,
(vxlan_addr_any(&vninode->remote_ip) ?
dst->remote_ip : vninode->remote_ip),
@@ -636,7 +633,7 @@ static void vxlan_vni_delete_group(struct vxlan_dev *vxlan,
vninode->vni, vninode->vni,
dst->remote_ifindex,
true);
- spin_unlock_bh(&vxlan->hash_lock[hash_index]);
+ spin_unlock_bh(&vxlan->hash_lock);
}
if (vxlan->dev->flags & IFF_UP) {
Powered by blists - more mailing lists