netdev - Re: [net-next 1/3] ch_ktls: Issue if connection offload fails

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: <20200922154443.17ed8b94@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Tue, 22 Sep 2020 15:44:43 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Rohit Maheshwari <rohitm@...lsio.com>
Cc:     netdev@...r.kernel.org, davem@...emloft.net, secdev@...lsio.com
Subject: Re: [net-next 1/3] ch_ktls: Issue if connection offload fails

On Tue, 22 Sep 2020 23:14:59 +0530 Rohit Maheshwari wrote:
> Since driver first return success to tls_dev_add, if req to HW is
> successful, but later if HW returns failure, that connection traffic
> fails permanently and connection status remains unknown to stack.
> 
> Fixes: 34aba2c45024 ("cxgb4/chcr : Register to tls add and del callback")
> Signed-off-by: Rohit Maheshwari <rohitm@...lsio.com>

>  #if IS_ENABLED(CONFIG_IPV6)
>  	} else {
> -		if (!sk->sk_ipv6only &&
> -		    ipv6_addr_type(&sk->sk_v6_daddr) == IPV6_ADDR_MAPPED) {
> -			tx_info->ip_family = AF_INET;
> -			ret = chcr_ktls_act_open_req(sk, tx_info, atid);
> -		} else {
> -			tx_info->ip_family = AF_INET6;
> -			ret = cxgb4_clip_get(tx_info->netdev,
> -					     (const u32 *)
> -					     &sk->sk_v6_rcv_saddr.s6_addr,
> -					     1);
> -			if (ret)
> -				goto out;
> -			ret = chcr_ktls_act_open_req6(sk, tx_info, atid);
> -		}
> +		ret = cxgb4_clip_get(tx_info->netdev, (const u32 *)
> +				     &sk->sk_v6_rcv_saddr,
> +				     1);
> +		if (ret)
> +			return ret;
> +		ret = chcr_ktls_act_open_req6(sk, tx_info, atid);

You removed the mapped socket handling which seems unrelated to the
rest of the patch.

> +	spin_lock(&tx_info->lock);
> +	tx_info->conn_up = true;
> +	spin_unlock(&tx_info->lock);

What's the context this lock is taken in? You seem to always do only
spin_lock(), does the control path not need to be _bh() or _irq()?