[net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Mugunthan V N]

Dual EMAC works with VLAN segregation of the ports, so default vlan needs
to be added in dual EMAC case else default vlan will be tagged for all
egress packets and vlan unaware switches/servers will drop packets
from the EVM.

Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
---
 drivers/net/ethernet/ti/cpsw.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index 5d5fec6..6133e4a 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -1201,8 +1201,7 @@ static int cpsw_ndo_open(struct net_device *ndev)
 	for_each_slave(priv, cpsw_slave_open, priv);
 
 	/* Add default VLAN */
-	if (!priv->data.dual_emac)
-		cpsw_add_default_vlan(priv);
+	cpsw_add_default_vlan(priv);
 
 	if (!cpsw_common_res_usage_state(priv)) {
 		/* setup tx dma to fixed prio and zero offset */
-- 
1.9.0

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Yegor Yefremov]

On Wed, Apr 9, 2014 at 8:04 AM, Mugunthan V N <mugunthanvnm@ti.com> wrote:
> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
> to be added in dual EMAC case else default vlan will be tagged for all
> egress packets and vlan unaware switches/servers will drop packets
> from the EVM.
>
> Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
> ---
>  drivers/net/ethernet/ti/cpsw.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
> index 5d5fec6..6133e4a 100644
> --- a/drivers/net/ethernet/ti/cpsw.c
> +++ b/drivers/net/ethernet/ti/cpsw.c
> @@ -1201,8 +1201,7 @@ static int cpsw_ndo_open(struct net_device *ndev)
>         for_each_slave(priv, cpsw_slave_open, priv);
>
>         /* Add default VLAN */
> -       if (!priv->data.dual_emac)
> -               cpsw_add_default_vlan(priv);
> +       cpsw_add_default_vlan(priv);
>
>         if (!cpsw_common_res_usage_state(priv)) {
>                 /* setup tx dma to fixed prio and zero offset */
> --
> 1.9.0

Tested-by: Yegor Yefremov <yegorslists@googlemail.com>

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[David Miller]

From: Mugunthan V N <mugunthanvnm@ti.com>
Date: Wed, 9 Apr 2014 11:34:40 +0530

> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
> to be added in dual EMAC case else default vlan will be tagged for all
> egress packets and vlan unaware switches/servers will drop packets
> from the EVM.
> 
> Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>

Applied, thanks.

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Yegor Yefremov]

On Wed, Apr 9, 2014 at 8:56 PM, David Miller <davem@davemloft.net> wrote:
> From: Mugunthan V N <mugunthanvnm@ti.com>
> Date: Wed, 9 Apr 2014 11:34:40 +0530
>
>> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
>> to be added in dual EMAC case else default vlan will be tagged for all
>> egress packets and vlan unaware switches/servers will drop packets
>> from the EVM.

Hi Mugunthan,

though this patch fixes drop packets issue, it introduces another
issue. Just connect both interfaces to one switch and configure
different subnets. As soon as you get a broadcast you have an endless
loop and network communication is not possible at all. Before this
patch you could operate as if you had really two independent NICs.

Yegor

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Mugunthan V N]

On Tuesday 15 April 2014 07:16 PM, Yegor Yefremov wrote:
> On Wed, Apr 9, 2014 at 8:56 PM, David Miller <davem@davemloft.net> wrote:
>> From: Mugunthan V N <mugunthanvnm@ti.com>
>> Date: Wed, 9 Apr 2014 11:34:40 +0530
>>
>>> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
>>> to be added in dual EMAC case else default vlan will be tagged for all
>>> egress packets and vlan unaware switches/servers will drop packets
>>> from the EVM.
> Hi Mugunthan,
>
> though this patch fixes drop packets issue, it introduces another
> issue. Just connect both interfaces to one switch and configure
> different subnets. As soon as you get a broadcast you have an endless
> loop and network communication is not possible at all. Before this
> patch you could operate as if you had really two independent NICs.
>
>

This patch doesn't add any multicast or unicast entry in the ALE table,
so there is no possible way that a broadcast packet is switched within
the two external ports, and this patch removes the encapsulation of the
default vlan id which is used for internal packet switching inside CPSW
switch from host and any packet received from external port will be
tagged with port vlan so internal switching will happen with that port
vlan-id which will be only that physical port and host port. so the
scenario which you mentioned cannot happen theoretically untill you
enable bridge between the dual ethernet interface and connect both to
the same interface.

Regards
Mugunthan V N

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Yegor Yefremov]

On Wed, Apr 16, 2014 at 7:57 AM, Mugunthan V N <mugunthanvnm@ti.com> wrote:
> On Tuesday 15 April 2014 07:16 PM, Yegor Yefremov wrote:
>> On Wed, Apr 9, 2014 at 8:56 PM, David Miller <davem@davemloft.net> wrote:
>>> From: Mugunthan V N <mugunthanvnm@ti.com>
>>> Date: Wed, 9 Apr 2014 11:34:40 +0530
>>>
>>>> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
>>>> to be added in dual EMAC case else default vlan will be tagged for all
>>>> egress packets and vlan unaware switches/servers will drop packets
>>>> from the EVM.
>> Hi Mugunthan,
>>
>> though this patch fixes drop packets issue, it introduces another
>> issue. Just connect both interfaces to one switch and configure
>> different subnets. As soon as you get a broadcast you have an endless
>> loop and network communication is not possible at all. Before this
>> patch you could operate as if you had really two independent NICs.
>>
>>
>
> This patch doesn't add any multicast or unicast entry in the ALE table,
> so there is no possible way that a broadcast packet is switched within
> the two external ports, and this patch removes the encapsulation of the
> default vlan id which is used for internal packet switching inside CPSW
> switch from host and any packet received from external port will be
> tagged with port vlan so internal switching will happen with that port
> vlan-id which will be only that physical port and host port. so the
> scenario which you mentioned cannot happen theoretically untill you
> enable bridge between the dual ethernet interface and connect both to
> the same interface.

Have you tried the described setup? My system is Buildroot on
am335x-evmsk with almost empty /etc/network/interfaces

# Configure Loopback
auto lo
iface lo inet loopback

No bridges:

# brctl show
bridge name     bridge id               STP enabled     interfaces

And following address setup made manually:

1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
group default
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UP group default qlen 1000
    link/ether 00:18:31:93:f6:49 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.234/24 brd 192.168.1.255 scope global eth0
       valid_lft forever preferred_lft forever
3: eth1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast
state DOWN group default qlen 1000
    link/ether 56:90:76:5b:b5:f3 brd ff:ff:ff:ff:ff:ff
    inet 192.168.4.234/24 brd 192.168.4.255 scope global eth1
       valid_lft forever preferred_lft forever
4: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group
default qlen 1000
    link/ether de:ad:be:ef:00:00 brd ff:ff:ff:ff:ff:ff

Routing table:

# ip route show
192.168.1.0/24 dev eth0  proto kernel  scope link  src 192.168.1.234
192.168.4.0/24 dev eth1  proto kernel  scope link  src 192.168.4.234

Same behavior on 3.14.0-rc4-12740-gcc09039-dirty and 3.15.0-rc1-00002-ge0f8779

As soon as insert the cable into eth1, my switch shows very high
activity and the whole LAN in the company is dead till I remove the
cable. I have no problems with the version before the fix. I don't
really know, what happens inside the driver, but could make the same
setup (best of all in a LAN with many Windows PCs, that send lots of
broadcast) and let me know if you see the same behavior.

Regards,
Yegor

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Yegor Yefremov]

On Wed, Apr 16, 2014 at 3:58 PM, Yegor Yefremov
<yegorslists@googlemail.com> wrote:
> On Wed, Apr 16, 2014 at 7:57 AM, Mugunthan V N <mugunthanvnm@ti.com> wrote:
>> On Tuesday 15 April 2014 07:16 PM, Yegor Yefremov wrote:
>>> On Wed, Apr 9, 2014 at 8:56 PM, David Miller <davem@davemloft.net> wrote:
>>>> From: Mugunthan V N <mugunthanvnm@ti.com>
>>>> Date: Wed, 9 Apr 2014 11:34:40 +0530
>>>>
>>>>> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
>>>>> to be added in dual EMAC case else default vlan will be tagged for all
>>>>> egress packets and vlan unaware switches/servers will drop packets
>>>>> from the EVM.
>>> Hi Mugunthan,
>>>
>>> though this patch fixes drop packets issue, it introduces another
>>> issue. Just connect both interfaces to one switch and configure
>>> different subnets. As soon as you get a broadcast you have an endless
>>> loop and network communication is not possible at all. Before this
>>> patch you could operate as if you had really two independent NICs.
>>>
>>>
>>
>> This patch doesn't add any multicast or unicast entry in the ALE table,
>> so there is no possible way that a broadcast packet is switched within
>> the two external ports, and this patch removes the encapsulation of the
>> default vlan id which is used for internal packet switching inside CPSW
>> switch from host and any packet received from external port will be
>> tagged with port vlan so internal switching will happen with that port
>> vlan-id which will be only that physical port and host port. so the
>> scenario which you mentioned cannot happen theoretically untill you
>> enable bridge between the dual ethernet interface and connect both to
>> the same interface.
>
> Have you tried the described setup? My system is Buildroot on
> am335x-evmsk with almost empty /etc/network/interfaces
>
> # Configure Loopback
> auto lo
> iface lo inet loopback
>
> No bridges:
>
> # brctl show
> bridge name     bridge id               STP enabled     interfaces
>
> And following address setup made manually:
>
> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
> group default
>     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>     inet 127.0.0.1/8 scope host lo
>        valid_lft forever preferred_lft forever
> 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
> state UP group default qlen 1000
>     link/ether 00:18:31:93:f6:49 brd ff:ff:ff:ff:ff:ff
>     inet 192.168.1.234/24 brd 192.168.1.255 scope global eth0
>        valid_lft forever preferred_lft forever
> 3: eth1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast
> state DOWN group default qlen 1000
>     link/ether 56:90:76:5b:b5:f3 brd ff:ff:ff:ff:ff:ff
>     inet 192.168.4.234/24 brd 192.168.4.255 scope global eth1
>        valid_lft forever preferred_lft forever
> 4: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group
> default qlen 1000
>     link/ether de:ad:be:ef:00:00 brd ff:ff:ff:ff:ff:ff
>
> Routing table:
>
> # ip route show
> 192.168.1.0/24 dev eth0  proto kernel  scope link  src 192.168.1.234
> 192.168.4.0/24 dev eth1  proto kernel  scope link  src 192.168.4.234
>
> Same behavior on 3.14.0-rc4-12740-gcc09039-dirty and 3.15.0-rc1-00002-ge0f8779
>
> As soon as insert the cable into eth1, my switch shows very high
> activity and the whole LAN in the company is dead till I remove the
> cable. I have no problems with the version before the fix. I don't
> really know, what happens inside the driver, but could make the same
> setup (best of all in a LAN with many Windows PCs, that send lots of
> broadcast) and let me know if you see the same behavior.

This bug seems to be in 3.12 too. See this thread
(http://e2e.ti.com/support/arm/sitara_arm/f/791/p/338292/1180954.aspx#1180954).

Thanks.

Yegor

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Mugunthan V N]

On Friday 02 May 2014 02:28 PM, Yegor Yefremov wrote:
> On Wed, Apr 16, 2014 at 3:58 PM, Yegor Yefremov
> <yegorslists@googlemail.com> wrote:
>> On Wed, Apr 16, 2014 at 7:57 AM, Mugunthan V N <mugunthanvnm@ti.com> wrote:
>>> On Tuesday 15 April 2014 07:16 PM, Yegor Yefremov wrote:
>>>> On Wed, Apr 9, 2014 at 8:56 PM, David Miller <davem@davemloft.net> wrote:
>>>>> From: Mugunthan V N <mugunthanvnm@ti.com>
>>>>> Date: Wed, 9 Apr 2014 11:34:40 +0530
>>>>>
>>>>>> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
>>>>>> to be added in dual EMAC case else default vlan will be tagged for all
>>>>>> egress packets and vlan unaware switches/servers will drop packets
>>>>>> from the EVM.
>>>> Hi Mugunthan,
>>>>
>>>> though this patch fixes drop packets issue, it introduces another
>>>> issue. Just connect both interfaces to one switch and configure
>>>> different subnets. As soon as you get a broadcast you have an endless
>>>> loop and network communication is not possible at all. Before this
>>>> patch you could operate as if you had really two independent NICs.
>>>>
>>>>
>>> This patch doesn't add any multicast or unicast entry in the ALE table,
>>> so there is no possible way that a broadcast packet is switched within
>>> the two external ports, and this patch removes the encapsulation of the
>>> default vlan id which is used for internal packet switching inside CPSW
>>> switch from host and any packet received from external port will be
>>> tagged with port vlan so internal switching will happen with that port
>>> vlan-id which will be only that physical port and host port. so the
>>> scenario which you mentioned cannot happen theoretically untill you
>>> enable bridge between the dual ethernet interface and connect both to
>>> the same interface.
>> Have you tried the described setup? My system is Buildroot on
>> am335x-evmsk with almost empty /etc/network/interfaces
>>
>> # Configure Loopback
>> auto lo
>> iface lo inet loopback
>>
>> No bridges:
>>
>> # brctl show
>> bridge name     bridge id               STP enabled     interfaces
>>
>> And following address setup made manually:
>>
>> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
>> group default
>>     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>>     inet 127.0.0.1/8 scope host lo
>>        valid_lft forever preferred_lft forever
>> 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
>> state UP group default qlen 1000
>>     link/ether 00:18:31:93:f6:49 brd ff:ff:ff:ff:ff:ff
>>     inet 192.168.1.234/24 brd 192.168.1.255 scope global eth0
>>        valid_lft forever preferred_lft forever
>> 3: eth1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast
>> state DOWN group default qlen 1000
>>     link/ether 56:90:76:5b:b5:f3 brd ff:ff:ff:ff:ff:ff
>>     inet 192.168.4.234/24 brd 192.168.4.255 scope global eth1
>>        valid_lft forever preferred_lft forever
>> 4: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group
>> default qlen 1000
>>     link/ether de:ad:be:ef:00:00 brd ff:ff:ff:ff:ff:ff
>>
>> Routing table:
>>
>> # ip route show
>> 192.168.1.0/24 dev eth0  proto kernel  scope link  src 192.168.1.234
>> 192.168.4.0/24 dev eth1  proto kernel  scope link  src 192.168.4.234
>>
>> Same behavior on 3.14.0-rc4-12740-gcc09039-dirty and 3.15.0-rc1-00002-ge0f8779
>>
>> As soon as insert the cable into eth1, my switch shows very high
>> activity and the whole LAN in the company is dead till I remove the
>> cable. I have no problems with the version before the fix. I don't
>> really know, what happens inside the driver, but could make the same
>> setup (best of all in a LAN with many Windows PCs, that send lots of
>> broadcast) and let me know if you see the same behavior.
> This bug seems to be in 3.12 too. See this thread
> (http://e2e.ti.com/support/arm/sitara_arm/f/791/p/338292/1180954.aspx#1180954).
>
> Thanks.
>
> Yegor

Will check this in this week.

Regards
Mugunthan V N

Re: [net PATCH 1/1] drivers: net: cpsw: Add default vlan for dual emac case also

[Yegor Yefremov]

On Mon, May 5, 2014 at 4:20 PM, Mugunthan V N <mugunthanvnm@ti.com> wrote:
> On Friday 02 May 2014 02:28 PM, Yegor Yefremov wrote:
>> On Wed, Apr 16, 2014 at 3:58 PM, Yegor Yefremov
>> <yegorslists@googlemail.com> wrote:
>>> On Wed, Apr 16, 2014 at 7:57 AM, Mugunthan V N <mugunthanvnm@ti.com> wrote:
>>>> On Tuesday 15 April 2014 07:16 PM, Yegor Yefremov wrote:
>>>>> On Wed, Apr 9, 2014 at 8:56 PM, David Miller <davem@davemloft.net> wrote:
>>>>>> From: Mugunthan V N <mugunthanvnm@ti.com>
>>>>>> Date: Wed, 9 Apr 2014 11:34:40 +0530
>>>>>>
>>>>>>> Dual EMAC works with VLAN segregation of the ports, so default vlan needs
>>>>>>> to be added in dual EMAC case else default vlan will be tagged for all
>>>>>>> egress packets and vlan unaware switches/servers will drop packets
>>>>>>> from the EVM.
>>>>> Hi Mugunthan,
>>>>>
>>>>> though this patch fixes drop packets issue, it introduces another
>>>>> issue. Just connect both interfaces to one switch and configure
>>>>> different subnets. As soon as you get a broadcast you have an endless
>>>>> loop and network communication is not possible at all. Before this
>>>>> patch you could operate as if you had really two independent NICs.
>>>>>
>>>>>
>>>> This patch doesn't add any multicast or unicast entry in the ALE table,
>>>> so there is no possible way that a broadcast packet is switched within
>>>> the two external ports, and this patch removes the encapsulation of the
>>>> default vlan id which is used for internal packet switching inside CPSW
>>>> switch from host and any packet received from external port will be
>>>> tagged with port vlan so internal switching will happen with that port
>>>> vlan-id which will be only that physical port and host port. so the
>>>> scenario which you mentioned cannot happen theoretically untill you
>>>> enable bridge between the dual ethernet interface and connect both to
>>>> the same interface.
>>> Have you tried the described setup? My system is Buildroot on
>>> am335x-evmsk with almost empty /etc/network/interfaces
>>>
>>> # Configure Loopback
>>> auto lo
>>> iface lo inet loopback
>>>
>>> No bridges:
>>>
>>> # brctl show
>>> bridge name     bridge id               STP enabled     interfaces
>>>
>>> And following address setup made manually:
>>>
>>> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
>>> group default
>>>     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>>>     inet 127.0.0.1/8 scope host lo
>>>        valid_lft forever preferred_lft forever
>>> 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
>>> state UP group default qlen 1000
>>>     link/ether 00:18:31:93:f6:49 brd ff:ff:ff:ff:ff:ff
>>>     inet 192.168.1.234/24 brd 192.168.1.255 scope global eth0
>>>        valid_lft forever preferred_lft forever
>>> 3: eth1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast
>>> state DOWN group default qlen 1000
>>>     link/ether 56:90:76:5b:b5:f3 brd ff:ff:ff:ff:ff:ff
>>>     inet 192.168.4.234/24 brd 192.168.4.255 scope global eth1
>>>        valid_lft forever preferred_lft forever
>>> 4: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group
>>> default qlen 1000
>>>     link/ether de:ad:be:ef:00:00 brd ff:ff:ff:ff:ff:ff
>>>
>>> Routing table:
>>>
>>> # ip route show
>>> 192.168.1.0/24 dev eth0  proto kernel  scope link  src 192.168.1.234
>>> 192.168.4.0/24 dev eth1  proto kernel  scope link  src 192.168.4.234
>>>
>>> Same behavior on 3.14.0-rc4-12740-gcc09039-dirty and 3.15.0-rc1-00002-ge0f8779
>>>
>>> As soon as insert the cable into eth1, my switch shows very high
>>> activity and the whole LAN in the company is dead till I remove the
>>> cable. I have no problems with the version before the fix. I don't
>>> really know, what happens inside the driver, but could make the same
>>> setup (best of all in a LAN with many Windows PCs, that send lots of
>>> broadcast) and let me know if you see the same behavior.
>> This bug seems to be in 3.12 too. See this thread
>> (http://e2e.ti.com/support/arm/sitara_arm/f/791/p/338292/1180954.aspx#1180954).
>>
>> Thanks.
>>
>> Yegor
>
> Will check this in this week.

Any news concerning this issue?

Thanks.

Yegor