netdev - Re: [PATCH net 4/4] bonding: change ipsec

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <169992e9-7f5c-4d4e-bf7a-15c64084d9b9@redhat.com>
Date: Tue, 30 Jul 2024 13:28:02 +0200
From: Paolo Abeni <pabeni@...hat.com>
To: Tariq Toukan <tariqt@...dia.com>, "David S. Miller"
 <davem@...emloft.net>, Jakub Kicinski <kuba@...nel.org>,
 Eric Dumazet <edumazet@...gle.com>, Jay Vosburgh <jv@...sburgh.net>,
 Andy Gospodarek <andy@...yhouse.net>
Cc: netdev@...r.kernel.org, Saeed Mahameed <saeedm@...dia.com>,
 Gal Pressman <gal@...dia.com>, Leon Romanovsky <leonro@...dia.com>,
 Jianbo Liu <jianbol@...dia.com>, Cosmin Ratiu <cratiu@...dia.com>
Subject: Re: [PATCH net 4/4] bonding: change ipsec_lock from spin lock to
 mutex

On 7/29/24 14:44, Tariq Toukan wrote:
> From: Jianbo Liu <jianbol@...dia.com>
> 
> In the cited commit, bond->ipsec_lock is added to protect ipsec_list,
> hence xdo_dev_state_add and xdo_dev_state_delete are called inside
> this lock. As ipsec_lock is spin lock and such xfrmdev ops may sleep,

minor nit: missing 'a' here ^^

> diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
> index 763d807be311..bced29813478 100644
> --- a/drivers/net/bonding/bond_main.c
> +++ b/drivers/net/bonding/bond_main.c
> @@ -439,38 +439,33 @@ static int bond_ipsec_add_sa(struct xfrm_state *xs,
>   	rcu_read_lock();
>   	bond = netdev_priv(bond_dev);
>   	slave = rcu_dereference(bond->curr_active_slave);

Is even this caller always under RTNL lock? if so it would be better 
replace  rcu_dereference() with rtnl_dereference() and drop the rcu 
lock, so it's clear that real_dev can't go away here.

Similar question for bond_ipsec_delete_sa, below.

> -	if (!slave) {
> -		rcu_read_unlock();
> +	real_dev = slave ? slave->dev : NULL;
> +	rcu_read_unlock();
> +	if (!real_dev)
>   		return -ENODEV;
> -	}
>   
> -	real_dev = slave->dev;
>   	if (!real_dev->xfrmdev_ops ||
>   	    !real_dev->xfrmdev_ops->xdo_dev_state_add ||
>   	    netif_is_bond_master(real_dev)) {
>   		NL_SET_ERR_MSG_MOD(extack, "Slave does not support ipsec offload");
> -		rcu_read_unlock();
>   		return -EINVAL;
>   	}
>   
>   	ipsec = kmalloc(sizeof(*ipsec), GFP_ATOMIC);

I guess at this point you can use GFP_KERNEL here.

[...]
> @@ -482,34 +477,44 @@ static void bond_ipsec_add_sa_all(struct bonding *bond)
>   	struct slave *slave;
>   
>   	rcu_read_lock();
> -	slave = rcu_dereference(bond->curr_active_slave);
> -	if (!slave)
> -		goto out;
> +	slave = rtnl_dereference(bond->curr_active_slave);
> +	real_dev = slave ? slave->dev : NULL;
> +	rcu_read_unlock();

You can drop the rcu read lock/unlock here.

[...]
> @@ -569,14 +574,13 @@ static void bond_ipsec_del_sa_all(struct bonding *bond)
>   	struct slave *slave;
>   
>   	rcu_read_lock();
> -	slave = rcu_dereference(bond->curr_active_slave);
> -	if (!slave) {
> -		rcu_read_unlock();
> +	slave = rtnl_dereference(bond->curr_active_slave);
> +	real_dev = slave ? slave->dev : NULL;
> +	rcu_read_unlock();

Same here.

Thanks,

Paolo