[PATCH] net: disable all other devices when enabling the new one

[PATCH] net: disable all other devices when enabling the new one

[Michael Grzeschik]

Some network devices need proper reinitialization before they can be
used. This is especially the case when using devices that share the same
driver.

Therefor the current approach, starting the open routine of an ether
device just once with the first call of send, runs those devices into
some unreliable behaviour.

This patch disables all devices that are not being used so they get
reinitialized with the next usage.

Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
---
 net/eth.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/net/eth.c b/net/eth.c
index 53d24baa16..0c26324d3f 100644
--- a/net/eth.c
+++ b/net/eth.c
@@ -217,6 +217,7 @@ static int eth_carrier_check(struct eth_device *edev, int force)
  */
 static int eth_check_open(struct eth_device *edev)
 {
+	struct eth_device *tmpedev;
 	int ret;
 
 	if (edev->active)
@@ -226,6 +227,9 @@ static int eth_check_open(struct eth_device *edev)
 	if (ret)
 		return ret;
 
+	for_each_netdev(tmpedev)
+		tmpedev->active = 0;
+
 	edev->active = 1;
 
 	return eth_carrier_check(edev, 1);
-- 
2.24.0


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] net: disable all other devices when enabling the new one

[Sascha Hauer]

On Fri, Nov 15, 2019 at 05:20:14PM +0100, Michael Grzeschik wrote:
> Some network devices need proper reinitialization before they can be
> used. This is especially the case when using devices that share the same
> driver.
> 
> Therefor the current approach, starting the open routine of an ether
> device just once with the first call of send, runs those devices into
> some unreliable behaviour.
> 
> This patch disables all devices that are not being used so they get
> reinitialized with the next usage.
> 
> Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
> ---
>  net/eth.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/net/eth.c b/net/eth.c
> index 53d24baa16..0c26324d3f 100644
> --- a/net/eth.c
> +++ b/net/eth.c
> @@ -217,6 +217,7 @@ static int eth_carrier_check(struct eth_device *edev, int force)
>   */
>  static int eth_check_open(struct eth_device *edev)
>  {
> +	struct eth_device *tmpedev;
>  	int ret;
>  
>  	if (edev->active)
> @@ -226,6 +227,9 @@ static int eth_check_open(struct eth_device *edev)
>  	if (ret)
>  		return ret;
>  
> +	for_each_netdev(tmpedev)
> +		tmpedev->active = 0;

This is wrong. barebox can handle multiple open network devices just
fine. The bug is in the cpsw driver like described in my response to
your cpsw patch.

Sascha


-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox