netdev - Re: [PATCH net 1/2] qede: fix to disable start_xmit functionality during self adapter test

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20210316145935.6544c29b@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Tue, 16 Mar 2021 14:59:35 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Bhaskar Upadhaya <bupadhaya@...vell.com>
Cc:     <netdev@...r.kernel.org>, <aelior@...vell.com>,
        <irusskikh@...vell.com>, <davem@...emloft.net>
Subject: Re: [PATCH net 1/2] qede: fix to disable start_xmit functionality
 during self adapter test

On Tue, 16 Mar 2021 11:34:09 -0700 Bhaskar Upadhaya wrote:
> start_xmit function should not be called during the execution of self
> adapter test, netif_tx_disable() gives this guarantee, since it takes
> the transmit queue lock while marking the queue stopped.  This will
> wait for the transmit function to complete before returning.
> 
> Fixes: 16f46bf054f8 ("qede: add implementation for internal loopback test.")
> Signed-off-by: Bhaskar Upadhaya <bupadhaya@...vell.com>
> Signed-off-by: Igor Russkikh <irusskikh@...vell.com>
> Signed-off-by: Ariel Elior <aelior@...vell.com>
> ---
>  drivers/net/ethernet/qlogic/qede/qede_ethtool.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
> index 1560ad3d9290..f9702cc7bc55 100644
> --- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
> +++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
> @@ -1611,7 +1611,7 @@ static int qede_selftest_run_loopback(struct qede_dev *edev, u32 loopback_mode)
>  		return -EINVAL;
>  	}
>  
> -	qede_netif_stop(edev);
> +	netif_tx_disable(edev->ndev);

But an interrupt can come in after and enable Tx again.
I think you should keep the qede_netif_stop() here instead of moving 
it down, no?

>  	/* Bring up the link in Loopback mode */
>  	memset(&link_params, 0, sizeof(link_params));
> @@ -1623,6 +1623,8 @@ static int qede_selftest_run_loopback(struct qede_dev *edev, u32 loopback_mode)
>  	/* Wait for loopback configuration to apply */
>  	msleep_interruptible(500);
>  
> +	qede_netif_stop(edev);
> +
>  	/* Setting max packet size to 1.5K to avoid data being split over
>  	 * multiple BDs in cases where MTU > PAGE_SIZE.
>  	 */