[OpenWrt-Devel] ar934x+ar8327v4

[David Hutchison]

I meant to say Could it be as simple as the switch just isn't fully
initialized? instead of *recognized*... the ar8216 driver definitely
recognizes it properly, and exposes the switch0 interface. I was
thinking perhaps something else had to happen during init for the
"eth0" interface.

-- Davey

On Fri, Dec 5, 2014 at 9:15 AM, David Hutchison <dhutchison at bluemesh.net> wrote:
> Hello,
>
> Yes the problem remains when enable_vlan is set to 0
>
> I don't think it's related to your changes either. I synced with trunk
> in hope that your changes made a difference with my problem. However
> that does not seem to be the case.
>
> I was hoping someone could point me in the right direction. Does this
> sound more like an issue with ag71xx since that is what exposes the
> "eth0" interface? Could it be as simple as the switch just isn't
> recognized? Perhaps the switch interconnect is no longer RGMII and
> should be GMII?
>
> If I knew where in those drivers to start looking, I could probably
> help fix this problem. I enjoy tinkering with the drivers, and
> learning how to fix them :)
>
> I was hoping you had some knowledge about how the interface becomes
> tied with the switch itself, and if there were any specifics for the
> ar8327 that must be set before the communication between the CPU +
> Switch Chip can happen.
>
> Felix, would you by chance have any ideas as to what to check next?
> The switch0 interface works fine, the eth0 interface tied to the
> switch. Does not.
>
> The other thing to note ( and i'm not sure if this is by design ).. I
> used to be able to do "swconfig dev eth0 show" and show the switch the
> eth0 was connected to. Now I must use "swconfig dev switch0 show" to
> configure the switch. When you do "swconfig dev eth0 show" it says it
> Failed to connect to the switch. In the past, I have always used eth0
> to configure the switch however that may have been incorrect.
>
> -- Davey
>
> On Thu, Dec 4, 2014 at 11:57 PM, Heiner Kallweit <hkallweit1 at gmail.com> wrote:
>> Am 05.12.2014 um 06:13 schrieb David Hutchison:
>>> Hi,
>>>
>>> I ran into an issue with the Mikrotik Routerboard 951G's switch chip.
>>> The new batch that we received use an updated ar8327 switch chip. The
>>> switch chip would not function properly so I decided to load trunk and
>>> utilize your patches.
>>>
>>> dmesg shows the following:
>>>
>>> switch0: Atheros AR8327 rev. 4 switch registered on ag71xx-mdio.0
>>> libphy: ag71xx_mdio: probed
>>> eth0: Atheros AG71xx at 0xb9000000, irq 4, mode:RGMII
>>> ag71xx ag71xx.0 eth0: connected to PHY at ag71xx-mdio.0:00
>>> [uid=004dd034, driver=Atheros AR8216/AR8236/AR8316]
>>>
>>> It appears the updated ar8216.c with your changes have successfully
>>> initialized the chip. In fact I can even use "swconfig dev switch0" to
>>> create vlans etc.
>>>
>>> If I create a port vlan ( eg. put ports 1 - 2 in vlan 1 ). I can plug
>>> ethernet cables into each port and push data from port 1 to port 2
>>> etc. It appears the switch driver is working correctly.
>>>
>>> What is broken is the interface handle "eth0". It exposes eth0 and
>>> ties it to the switch correctly ( as you can see from the above dmesg
>>> output ). I then assign an IP address to eth0, however when I try to
>>> ping a device connected to the switch. I get no response. There is no
>>> arp traffic either. It's like data from the "eth0" handle doesn't know
>>> how to route through the switch.
>>>
>>> It appears like the switch hardware is working ( since port 1 can
>>> communicate to port 2 ). However when "eth0" attempts to communicate
>>> to a device on port 1, it doesn't work. The device on port 1 cannot
>>> communicate with the "eth0" interface either.
>>>
>>> I think your patches are specifically for the switch only. If that's
>>> the case, then they are working. However, with your knowledge of
>>> ar8327 could you point me into the direction of fixing the "eth0"
>>> interface communicating to the ar8327 switch properly? Is there
>>> something in ag71xx that I should be looking at? Could the ar8327
>>> switch not be initializing properly? The probe is definitely finding
>>> *AR8327* so that seems to be OK. I am guessing this is something
>>> specific with the ar924x CPU + ar8327 switch chip. I'm open to any
>>> suggestions :-)
>>>
>>> -- Davey
>> The recent changes to the AR8216 driver are mainly cleanups with
>> no functional change. Also your dmesg output looks good.
>> So at a first glance I don't think it's something with the driver.
>>
>> At first your /etc/config/network would be helpful.
>>
>> Does your problem also happen if vlans are switched off
>> (enable_vlan set to 0) ?
>>
>> Heiner
>>
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel